Gesture_Recognition_code.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "jKnFk-08CjDy"
   },
   "source": [
    "# Gesture Recognition\n",
    "\n",
    "In this project, you are going to build a 3D Conv model that will be able to predict the 5 gestures correctly. Please import the following libraries to get started."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "KxstHs2bCjDz"
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import os\n",
    "import imageio\n",
    "from skimage.transform import resize, rescale\n",
    "import datetime\n",
    "import os\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from keras.models import Sequential, Model\n",
    "from keras.layers import Dense, Dropout, GRU, LSTM, Flatten, TimeDistributed, Flatten, BatchNormalization, Activation\n",
    "from keras.layers.convolutional import Conv3D, MaxPooling3D, Conv2D, MaxPooling2D\n",
    "from keras.callbacks import ModelCheckpoint, ReduceLROnPlateau\n",
    "from keras.regularizers import l2\n",
    "from keras import optimizers"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "dBgO9n9SCjD3"
   },
   "source": [
    "We set the random seed so that the results don't vary drastically."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "lMJqGvgLCjD4",
    "outputId": "e7e785f8-f35f-4d8f-94ab-d8987c7b8ea3"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.7.0\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "print(tf.__version__)\n",
    "np.random.seed(30)\n",
    "import random as rn\n",
    "rn.seed(30)\n",
    "from keras import backend as K\n",
    "tf.random.set_seed(30)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "fVVm2F2rCjD8"
   },
   "source": [
    "In this block, you read the folder names for training and validation. You also set the `batch_size` here. Note that you set the batch size in such a way that you are able to use the GPU in full capacity. You keep increasing the batch size until the machine throws an error."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "-Ss1MnP9CjD8"
   },
   "outputs": [],
   "source": [
    "train_doc = np.random.permutation(open('/home/datasets/Project_data/train.csv').readlines())\n",
    "val_doc = np.random.permutation(open('/home/datasets/Project_data/val.csv').readlines())\n",
    "num_classes = 5  #number of softmax classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "id": "tCVJko_xCjD_",
    "outputId": "a17ef715-290f-44e9-8702-14b1998f7f5c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# training sequences = 663\n",
      "# validation sequences = 100\n"
     ]
    }
   ],
   "source": [
    "curr_dt_time = datetime.datetime.now()\n",
    "train_path = '/home/datasets/Project_data/train'\n",
    "val_path = '/home/datasets/Project_data/val'\n",
    "num_train_sequences = len(train_doc)\n",
    "print('# training sequences =', num_train_sequences)\n",
    "num_val_sequences = len(val_doc)\n",
    "print('# validation sequences =', num_val_sequences)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "l2qVMqXK7EFp"
   },
   "source": [
    "## Common Utils"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "9XID1rKr4Jwh"
   },
   "source": [
    "Common utility to plot loss and accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "R4bgSpPQIrgn"
   },
   "outputs": [],
   "source": [
    "def plot_loss_accuracy(history):\n",
    "    # list all data in history\n",
    "    print(history.history.keys())\n",
    "    # summarize history for accuracy\n",
    "    plt.plot(history.history['categorical_accuracy'])\n",
    "    plt.plot(history.history['val_categorical_accuracy'])\n",
    "    plt.title('model accuracy')\n",
    "    plt.ylabel('accuracy')\n",
    "    plt.xlabel('epoch')\n",
    "    plt.legend(['train', 'test'], loc='upper left')\n",
    "    plt.show()\n",
    "    # summarize history for loss\n",
    "    plt.plot(history.history['loss'])\n",
    "    plt.plot(history.history['val_loss'])\n",
    "    plt.title('model loss')\n",
    "    plt.ylabel('loss')\n",
    "    plt.xlabel('epoch')\n",
    "    plt.legend(['train', 'test'], loc='upper left')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "drEZFmAG4Sbd"
   },
   "source": [
    "Common utility to calculate steps"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "p5ynijKOJhrN"
   },
   "outputs": [],
   "source": [
    "def calculate_steps(num_train_sequences, num_val_sequences, batch_size):\n",
    "    if (num_train_sequences%batch_size) == 0:\n",
    "        steps_per_epoch = int(num_train_sequences/batch_size)\n",
    "    else:\n",
    "        steps_per_epoch = (num_train_sequences//batch_size) + 1\n",
    "\n",
    "    if (num_val_sequences%batch_size) == 0:\n",
    "        validation_steps = int(num_val_sequences/batch_size)\n",
    "    else:\n",
    "        validation_steps = (num_val_sequences//batch_size) + 1\n",
    "\n",
    "    return steps_per_epoch,validation_steps"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "AcwwPnM84ZSm"
   },
   "source": [
    "Common utility for callbacks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "bU94bF80PFEx"
   },
   "outputs": [],
   "source": [
    "def model_callbacks(folder_name):\n",
    "    model_name = str(folder_name) + '_' + str(curr_dt_time).replace(' ','').replace(':','_') + '/'\n",
    "   \n",
    "    if not os.path.exists(model_name):\n",
    "        os.mkdir(model_name)\n",
    "\n",
    "    filepath = model_name + 'model-{epoch:05d}-{loss:.5f}-{categorical_accuracy:.5f}-{val_loss:.5f}-{val_categorical_accuracy:.5f}.h5'\n",
    "\n",
    "    checkpoint = ModelCheckpoint(filepath, monitor='val_loss', verbose=1, save_best_only=True, save_weights_only=False, mode='auto', period=1)\n",
    "\n",
    "    LR = ReduceLROnPlateau(monitor='val_loss', factor=0.1, patience=3, min_lr=0.0001, cooldown=1, verbose=1) # write the REducelronplateau code here\n",
    "\n",
    "    return [checkpoint, LR]    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "KnjCHjHjCjEC"
   },
   "source": [
    "## Generator\n",
    "This is one of the most important part of the code. The overall structure of the generator has been given. In the generator, you are going to preprocess the images as you have images of 2 different dimensions as well as create a batch of video frames. You have to experiment with `img_idx`, `y`,`z` and normalization such that you get high accuracy."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 573
    },
    "colab_type": "code",
    "id": "duZ2bmi0CjED",
    "outputId": "593c98d1-7987-44c2-a25d-9e84deebb4e3"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.8/dist-packages/keras_preprocessing/image/image_data_generator.py:337: UserWarning: This ImageDataGenerator specifies `zca_whitening`, which overrides setting of `featurewise_center`.\n",
      "  warnings.warn('This ImageDataGenerator specifies '\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "#from keras.preprocessing.image import smart_resize\n",
    "\n",
    "datagen = ImageDataGenerator(\n",
    "    zoom_range=0.1,\n",
    "    zca_whitening=True,\n",
    "    width_shift_range=0.1,\n",
    "    height_shift_range=0.1)\n",
    "\n",
    "## Test code start\n",
    "#360 by 360\n",
    "image1 = imageio.imread(\"/home/datasets/Project_data/val/WIN_20180907_15_30_06_Pro_Right Swipe_new/WIN_20180907_15_30_06_Pro_00029.png\")\n",
    "#160 by 120\n",
    "image2 = imageio.imread(\"/home/datasets/Project_data/val/WIN_20180926_17_56_52_Pro_Right_Swipe_new/WIN_20180926_17_56_52_Pro_00029.png\")\n",
    "#crop the images and resize them. Note that the images are of 2 different shape \n",
    "#and the conv3D will throw error if the inputs in a batch have different shapes                    \n",
    "#print(image1.shape, image2.shape)\n",
    "image1 = resize(image1, (120, 120), anti_aliasing=True)\n",
    "image1 = datagen.random_transform(image1)                   \n",
    "plt.imshow(image1)\n",
    "plt.show()\n",
    "image2 = resize(image2, (120, 120), anti_aliasing=True)\n",
    "image2 = datagen.random_transform(image2)                 \n",
    "plt.imshow(image2)\n",
    "plt.show()\n",
    "## Test code end\n",
    "\n",
    "def generate(source_path, folder_list, batch_size, num_images=20, height=120, width=120, augment=False):\n",
    "    #print( 'Source path = ', source_path, '; batch size =', batch_size)\n",
    "    total_frames = 30\n",
    "\n",
    "    while True:\n",
    "        t = np.random.permutation(folder_list)\n",
    "        num_batches = len(t)//batch_size # calculate the number of batches\n",
    "        #img_idx = sorted(rn.sample(range(0, total_frames), num_images)) #create a list of image numbers you want to use for a particular video\n",
    "        img_idx = np.round(np.linspace(0,total_frames-1,num_images)).astype(int) #create a list of image numbers you want to use for a particular video\n",
    "\n",
    "        for batch in range(num_batches): # we iterate over the number of batches\n",
    "            batch_data   = np.zeros((batch_size, num_images, height, width, 3)) # 3 is the number of channels RGB\n",
    "            batch_labels = np.zeros((batch_size, 5)) # batch_labels is the one hot representation of the output           \n",
    "\n",
    "            for folder in range(batch_size): # iterate over the batch_size\n",
    "                imgs = os.listdir(source_path+'/'+ t[folder + (batch*batch_size)].split(';')[0]) # read all the images in the folder\n",
    "                for idx,item in enumerate(img_idx): #  Iterate over the frames/images of a folder to read them in\n",
    "                    image = imageio.imread(source_path+'/'+ t[folder + (batch*batch_size)].strip().split(';')[0]+'/'+imgs[item]).astype(np.float32)\n",
    "\n",
    "                    #crop the images and resize them. Note that the images are of 2 different shape \n",
    "                    #and the conv3D will throw error if the inputs in a batch have different shapes                    \n",
    "                    h, w, c = image.shape\n",
    "                    image = resize(image, (height, width), anti_aliasing=True)\n",
    "\n",
    "                    # Randomly transform few images of few folders; note that folders are randomly shuffled in each epoch too\n",
    "                    if augment:\n",
    "                        if np.random.randn() > 0:\n",
    "                            image = datagen.random_transform(image)                   \n",
    "\n",
    "                    batch_data[folder,idx,:,:,0] = (image[...,0])/255 #normalise and feed in the image\n",
    "                    batch_data[folder,idx,:,:,1] = (image[...,1])/255 #normalise and feed in the image\n",
    "                    batch_data[folder,idx,:,:,2] = (image[...,2])/255 #normalise and feed in the image                  \n",
    "                    \n",
    "                batch_labels[folder, int(t[folder + (batch*batch_size)].strip().split(';')[2])] = 1\n",
    "\n",
    "            yield batch_data, batch_labels #you yield the batch_data and the batch_labels, remember what does yield do\n",
    "        \n",
    "        # write the code for the remaining data points which are left after full batches\n",
    "        if (len(t)%batch_size):\n",
    "            remaining_batch_size = len(t)%batch_size\n",
    "            batch_data   = np.zeros((remaining_batch_size, num_images, height, width,3)) # 3 is the number of channels RGB\n",
    "            batch_labels = np.zeros((remaining_batch_size,5)) # batch_labels is the one hot representation of the output           \n",
    "\n",
    "            for folder in range(remaining_batch_size): # iterate over the batch_size\n",
    "                imgs = os.listdir(source_path+'/'+ t[folder + (num_batches*batch_size)].split(';')[0]) # read all the images in the folder\n",
    "                for idx,item in enumerate(img_idx): #  Iterate iver the frames/images of a folder to read them in\n",
    "                    image = imageio.imread(source_path+'/'+ t[folder + (num_batches*batch_size)].strip().split(';')[0]+'/'+imgs[item]).astype(np.float32)\n",
    "                \n",
    "                    #crop the images and resize them. Note that the images are of 2 different shape \n",
    "                    #and the conv3D will throw error if the inputs in a batch have different shapes\n",
    "                    h, w, c = image.shape\n",
    "                    image = resize(image, (height, width), anti_aliasing=True)                          \n",
    "                    \n",
    "                    # Randomly transform few images of few folders; note that folders are randomly shuffled in each epoch too\n",
    "                    if augment:\n",
    "                        if np.random.randn() > 0:\n",
    "                            image = datagen.random_transform(image)                   \n",
    "\n",
    "                    batch_data[folder,idx,:,:,0] = (image[...,0])/255 #normalise and feed in the image\n",
    "                    batch_data[folder,idx,:,:,1] = (image[...,1])/255 #normalise and feed in the image\n",
    "                    batch_data[folder,idx,:,:,2] = (image[...,2])/255 #normalise and feed in the image\n",
    "                   \n",
    "                batch_labels[folder, int(t[folder + (num_batches*batch_size)].strip().split(';')[2])] = 1\n",
    "\n",
    "            yield batch_data, batch_labels #you yield the batch_data and the batch_labels, remember what does yield do\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "vyePp_mjCjEF"
   },
   "source": [
    "Note here that a video is represented above in the generator as (number of images, height, width, number of channels). Take this into consideration while creating the model architecture."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "sArsmY1mCjEG"
   },
   "source": [
    "## Model\n",
    "Here you make the model using different functionalities that Keras provides. Remember to use `Conv3D` and `MaxPooling3D` and not `Conv2D` and `Maxpooling2D` for a 3D convolution model. You would want to use `TimeDistributed` while building a Conv2D + RNN model. Also remember that the last layer is the softmax. Design the network in such a way that the model is able to give good accuracy on the least number of parameters so that it can fit in the memory of the webcam."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "1wzYeWUVCjEH"
   },
   "outputs": [],
   "source": [
    "from keras.models import Sequential, Model\n",
    "from keras.layers import Dense, Dropout, GRU, LSTM, Flatten, TimeDistributed, Flatten, BatchNormalization, Activation\n",
    "from keras.layers.convolutional import Conv3D, MaxPooling3D, Conv2D, MaxPooling2D\n",
    "from keras.callbacks import ModelCheckpoint, ReduceLROnPlateau\n",
    "from keras.regularizers import l2\n",
    "from keras import optimizers"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "vnvR4_KPCjEK"
   },
   "source": [
    "### Conv3d Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "DtvSVBWACjEN"
   },
   "outputs": [],
   "source": [
    "# Common utility to create conv 3D models\n",
    "def conv3D(conv_filters=(16, 32, 64, 128), dense_nodes=(256,128), dropout=0.25, num_images=20, height=120, width=120):\n",
    "\n",
    "    model = Sequential()\n",
    "\n",
    "    model.add(Conv3D(conv_filters[0], (3, 3, 3), padding='same', input_shape=(num_images, height, width, 3)))\n",
    "    model.add(Activation('relu'))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(MaxPooling3D(pool_size=(2, 2, 2)))\n",
    "\n",
    "    model.add(Conv3D(conv_filters[1], (3, 3, 3), padding='same'))\n",
    "    model.add(Activation('relu'))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(MaxPooling3D(pool_size=(2, 2, 2)))\n",
    "    \n",
    "    model.add(Conv3D(conv_filters[2], (3, 3, 3), padding='same'))\n",
    "    model.add(Activation('relu'))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(MaxPooling3D(pool_size=(2, 2, 2)))\n",
    "    \n",
    "    model.add(Conv3D(conv_filters[3], (3, 3, 3), padding='same'))    \n",
    "    model.add(Activation('relu'))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(MaxPooling3D(pool_size=(2, 2, 2)))\n",
    "    \n",
    "    model.add(Flatten())\n",
    "    model.add(Dense(dense_nodes[0]))\n",
    "    model.add(Activation('relu'))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(Dropout(dropout))\n",
    "\n",
    "    model.add(Dense(dense_nodes[1]))\n",
    "    model.add(Activation('relu'))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(Dropout(dropout))\n",
    "\n",
    "    model.add(Dense(num_classes))\n",
    "    model.add(Activation('softmax'))\n",
    "\n",
    "    opt = tf.keras.optimizers.Adam() #write your optimizer\n",
    "    model.compile(opt, loss='categorical_crossentropy', metrics=['categorical_accuracy'])    \n",
    "    \n",
    "    return model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "w0sfM-uWJ96m"
   },
   "source": [
    "#### Model 1: Conv 3D Model with 30 frames per video (16, 32, 64, 128 filters conv 3D layers + 256 dense layer + 128 dense layer + image size 120 by 120)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "KbLRWguYCjES",
    "outputId": "77319f4b-dfce-4132-84a2-6fa2d49a22e9"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# epochs = 25\n",
      "# batch size = 10\n",
      "# num_frames per video = 30\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-08-29 07:28:41.805433: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:39] Overriding allow_growth setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.\n",
      "2023-08-29 07:28:41.805495: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1525] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 14796 MB memory:  -> device: 0, name: Quadro RTX 5000, pci bus id: 0000:1c:00.0, compute capability: 7.5\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv3d (Conv3D)             (None, 30, 120, 120, 16)  1312      \n",
      "                                                                 \n",
      " activation (Activation)     (None, 30, 120, 120, 16)  0         \n",
      "                                                                 \n",
      " batch_normalization (BatchN  (None, 30, 120, 120, 16)  64       \n",
      " ormalization)                                                   \n",
      "                                                                 \n",
      " max_pooling3d (MaxPooling3D  (None, 15, 60, 60, 16)   0         \n",
      " )                                                               \n",
      "                                                                 \n",
      " conv3d_1 (Conv3D)           (None, 15, 60, 60, 32)    13856     \n",
      "                                                                 \n",
      " activation_1 (Activation)   (None, 15, 60, 60, 32)    0         \n",
      "                                                                 \n",
      " batch_normalization_1 (Batc  (None, 15, 60, 60, 32)   128       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " max_pooling3d_1 (MaxPooling  (None, 7, 30, 30, 32)    0         \n",
      " 3D)                                                             \n",
      "                                                                 \n",
      " conv3d_2 (Conv3D)           (None, 7, 30, 30, 64)     55360     \n",
      "                                                                 \n",
      " activation_2 (Activation)   (None, 7, 30, 30, 64)     0         \n",
      "                                                                 \n",
      " batch_normalization_2 (Batc  (None, 7, 30, 30, 64)    256       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " max_pooling3d_2 (MaxPooling  (None, 3, 15, 15, 64)    0         \n",
      " 3D)                                                             \n",
      "                                                                 \n",
      " conv3d_3 (Conv3D)           (None, 3, 15, 15, 128)    221312    \n",
      "                                                                 \n",
      " activation_3 (Activation)   (None, 3, 15, 15, 128)    0         \n",
      "                                                                 \n",
      " batch_normalization_3 (Batc  (None, 3, 15, 15, 128)   512       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " max_pooling3d_3 (MaxPooling  (None, 1, 7, 7, 128)     0         \n",
      " 3D)                                                             \n",
      "                                                                 \n",
      " flatten (Flatten)           (None, 6272)              0         \n",
      "                                                                 \n",
      " dense (Dense)               (None, 256)               1605888   \n",
      "                                                                 \n",
      " activation_4 (Activation)   (None, 256)               0         \n",
      "                                                                 \n",
      " batch_normalization_4 (Batc  (None, 256)              1024      \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " dropout (Dropout)           (None, 256)               0         \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 128)               32896     \n",
      "                                                                 \n",
      " activation_5 (Activation)   (None, 128)               0         \n",
      "                                                                 \n",
      " batch_normalization_5 (Batc  (None, 128)              512       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " dropout_1 (Dropout)         (None, 128)               0         \n",
      "                                                                 \n",
      " dense_2 (Dense)             (None, 5)                 645       \n",
      "                                                                 \n",
      " activation_6 (Activation)   (None, 5)                 0         \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 1,933,765\n",
      "Trainable params: 1,932,517\n",
      "Non-trainable params: 1,248\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "num_epochs = 20 # choose the number of epochs\n",
    "print ('# epochs =', num_epochs)\n",
    "batch_size = 10  #experiment with the batch size\n",
    "print ('# batch size =', batch_size)\n",
    "num_frames = 30\n",
    "print ('# num_frames per video =', num_frames)\n",
    "height = 120\n",
    "width = 120\n",
    "\n",
    "model = conv3D(num_images=num_frames)\n",
    "\n",
    "print (model.summary())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "VYLIR8rpCjER"
   },
   "source": [
    "The `steps_per_epoch` and `validation_steps` are used by `fit_generator` to decide the number of next() calls it need to make.\n",
    "Let us now fit the model. This will start training the model and with the help of the checkpoints, you'll be able to save the model at the end of each epoch."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "OAtEhUAtCjEU"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:`period` argument is deprecated. Please use `save_freq` to specify the frequency in number of batches seen.\n"
     ]
    }
   ],
   "source": [
    "callbacks_list = model_callbacks(\"model_conv3D_1\")\n",
    "\n",
    "steps_per_epoch, validation_steps = calculate_steps(num_train_sequences, num_val_sequences, batch_size)\n",
    "\n",
    "train_generator = generate(train_path, train_doc, batch_size, num_images=num_frames)\n",
    "\n",
    "val_generator   = generate(val_path, val_doc, batch_size, num_images=num_frames)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "NRBJ9DyOQyQj",
    "outputId": "2e469843-e403-4f6f-cb69-fe7b5e96aa48"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/25\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-08-28 12:47:22.561028: I tensorflow/stream_executor/cuda/cuda_dnn.cc:377] Loaded cuDNN version 8302\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "67/67 [==============================] - ETA: 0s - loss: 1.6461 - categorical_accuracy: 0.4238\n",
      "Epoch 00001: val_loss improved from inf to 5.75922, saving model to model_conv3D_1_2023-08-2812_46_28.519095/model-00001-1.64605-0.42383-5.75922-0.16000.h5\n",
      "67/67 [==============================] - 164s 2s/step - loss: 1.6461 - categorical_accuracy: 0.4238 - val_loss: 5.7592 - val_categorical_accuracy: 0.1600 - lr: 0.0010\n",
      "Epoch 2/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 1.1180 - categorical_accuracy: 0.5445\n",
      "Epoch 00002: val_loss did not improve from 5.75922\n",
      "67/67 [==============================] - 161s 2s/step - loss: 1.1180 - categorical_accuracy: 0.5445 - val_loss: 7.8871 - val_categorical_accuracy: 0.1900 - lr: 0.0010\n",
      "Epoch 3/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.8761 - categorical_accuracy: 0.6471\n",
      "Epoch 00003: val_loss did not improve from 5.75922\n",
      "67/67 [==============================] - 159s 2s/step - loss: 0.8761 - categorical_accuracy: 0.6471 - val_loss: 7.3735 - val_categorical_accuracy: 0.2200 - lr: 0.0010\n",
      "Epoch 4/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.7299 - categorical_accuracy: 0.7240\n",
      "Epoch 00004: val_loss did not improve from 5.75922\n",
      "\n",
      "Epoch 00004: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.\n",
      "67/67 [==============================] - 161s 2s/step - loss: 0.7299 - categorical_accuracy: 0.7240 - val_loss: 8.7108 - val_categorical_accuracy: 0.1900 - lr: 0.0010\n",
      "Epoch 5/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.5783 - categorical_accuracy: 0.7964\n",
      "Epoch 00005: val_loss did not improve from 5.75922\n",
      "67/67 [==============================] - 159s 2s/step - loss: 0.5783 - categorical_accuracy: 0.7964 - val_loss: 5.8878 - val_categorical_accuracy: 0.2400 - lr: 1.0000e-04\n",
      "Epoch 6/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.4890 - categorical_accuracy: 0.8190\n",
      "Epoch 00006: val_loss improved from 5.75922 to 4.26195, saving model to model_conv3D_1_2023-08-2812_46_28.519095/model-00006-0.48905-0.81900-4.26195-0.24000.h5\n",
      "67/67 [==============================] - 160s 2s/step - loss: 0.4890 - categorical_accuracy: 0.8190 - val_loss: 4.2619 - val_categorical_accuracy: 0.2400 - lr: 1.0000e-04\n",
      "Epoch 7/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.4393 - categorical_accuracy: 0.8386\n",
      "Epoch 00007: val_loss improved from 4.26195 to 2.86616, saving model to model_conv3D_1_2023-08-2812_46_28.519095/model-00007-0.43930-0.83861-2.86616-0.35000.h5\n",
      "67/67 [==============================] - 159s 2s/step - loss: 0.4393 - categorical_accuracy: 0.8386 - val_loss: 2.8662 - val_categorical_accuracy: 0.3500 - lr: 1.0000e-04\n",
      "Epoch 8/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3900 - categorical_accuracy: 0.8567\n",
      "Epoch 00008: val_loss improved from 2.86616 to 1.85930, saving model to model_conv3D_1_2023-08-2812_46_28.519095/model-00008-0.39004-0.85671-1.85930-0.52000.h5\n",
      "67/67 [==============================] - 161s 2s/step - loss: 0.3900 - categorical_accuracy: 0.8567 - val_loss: 1.8593 - val_categorical_accuracy: 0.5200 - lr: 1.0000e-04\n",
      "Epoch 9/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3894 - categorical_accuracy: 0.8552\n",
      "Epoch 00009: val_loss improved from 1.85930 to 0.97804, saving model to model_conv3D_1_2023-08-2812_46_28.519095/model-00009-0.38940-0.85520-0.97804-0.66000.h5\n",
      "67/67 [==============================] - 159s 2s/step - loss: 0.3894 - categorical_accuracy: 0.8552 - val_loss: 0.9780 - val_categorical_accuracy: 0.6600 - lr: 1.0000e-04\n",
      "Epoch 10/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3077 - categorical_accuracy: 0.8869\n",
      "Epoch 00010: val_loss improved from 0.97804 to 0.73762, saving model to model_conv3D_1_2023-08-2812_46_28.519095/model-00010-0.30769-0.88688-0.73762-0.74000.h5\n",
      "67/67 [==============================] - 160s 2s/step - loss: 0.3077 - categorical_accuracy: 0.8869 - val_loss: 0.7376 - val_categorical_accuracy: 0.7400 - lr: 1.0000e-04\n",
      "Epoch 11/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3097 - categorical_accuracy: 0.8944\n",
      "Epoch 00011: val_loss improved from 0.73762 to 0.65570, saving model to model_conv3D_1_2023-08-2812_46_28.519095/model-00011-0.30969-0.89442-0.65570-0.76000.h5\n",
      "67/67 [==============================] - 160s 2s/step - loss: 0.3097 - categorical_accuracy: 0.8944 - val_loss: 0.6557 - val_categorical_accuracy: 0.7600 - lr: 1.0000e-04\n",
      "Epoch 12/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2905 - categorical_accuracy: 0.8869\n",
      "Epoch 00012: val_loss did not improve from 0.65570\n",
      "67/67 [==============================] - 160s 2s/step - loss: 0.2905 - categorical_accuracy: 0.8869 - val_loss: 0.6841 - val_categorical_accuracy: 0.8100 - lr: 1.0000e-04\n",
      "Epoch 13/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2835 - categorical_accuracy: 0.8989\n",
      "Epoch 00013: val_loss improved from 0.65570 to 0.36275, saving model to model_conv3D_1_2023-08-2812_46_28.519095/model-00013-0.28355-0.89894-0.36275-0.86000.h5\n",
      "67/67 [==============================] - 160s 2s/step - loss: 0.2835 - categorical_accuracy: 0.8989 - val_loss: 0.3628 - val_categorical_accuracy: 0.8600 - lr: 1.0000e-04\n",
      "Epoch 14/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2309 - categorical_accuracy: 0.9321\n",
      "Epoch 00014: val_loss did not improve from 0.36275\n",
      "67/67 [==============================] - 162s 2s/step - loss: 0.2309 - categorical_accuracy: 0.9321 - val_loss: 0.4666 - val_categorical_accuracy: 0.8300 - lr: 1.0000e-04\n",
      "Epoch 15/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2560 - categorical_accuracy: 0.9095\n",
      "Epoch 00015: val_loss did not improve from 0.36275\n",
      "67/67 [==============================] - 162s 2s/step - loss: 0.2560 - categorical_accuracy: 0.9095 - val_loss: 0.5905 - val_categorical_accuracy: 0.8200 - lr: 1.0000e-04\n",
      "Epoch 16/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1891 - categorical_accuracy: 0.9351\n",
      "Epoch 00016: val_loss did not improve from 0.36275\n",
      "\n",
      "Epoch 00016: ReduceLROnPlateau reducing learning rate to 0.0001.\n",
      "67/67 [==============================] - 162s 2s/step - loss: 0.1891 - categorical_accuracy: 0.9351 - val_loss: 0.5165 - val_categorical_accuracy: 0.8200 - lr: 1.0000e-04\n",
      "Epoch 17/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1616 - categorical_accuracy: 0.9457\n",
      "Epoch 00017: val_loss did not improve from 0.36275\n",
      "67/67 [==============================] - 161s 2s/step - loss: 0.1616 - categorical_accuracy: 0.9457 - val_loss: 0.5417 - val_categorical_accuracy: 0.8100 - lr: 1.0000e-04\n",
      "Epoch 18/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1425 - categorical_accuracy: 0.9532\n",
      "Epoch 00018: val_loss did not improve from 0.36275\n",
      "67/67 [==============================] - 162s 2s/step - loss: 0.1425 - categorical_accuracy: 0.9532 - val_loss: 0.3719 - val_categorical_accuracy: 0.8400 - lr: 1.0000e-04\n",
      "Epoch 19/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1410 - categorical_accuracy: 0.9593\n",
      "Epoch 00019: val_loss did not improve from 0.36275\n",
      "67/67 [==============================] - 162s 2s/step - loss: 0.1410 - categorical_accuracy: 0.9593 - val_loss: 0.5456 - val_categorical_accuracy: 0.8500 - lr: 1.0000e-04\n",
      "Epoch 20/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1812 - categorical_accuracy: 0.9412\n",
      "Epoch 00020: val_loss did not improve from 0.36275\n",
      "67/67 [==============================] - 162s 2s/step - loss: 0.1812 - categorical_accuracy: 0.9412 - val_loss: 0.5401 - val_categorical_accuracy: 0.8400 - lr: 1.0000e-04\n",
      "Epoch 21/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1193 - categorical_accuracy: 0.9638\n",
      "Epoch 00021: val_loss did not improve from 0.36275\n",
      "67/67 [==============================] - 162s 2s/step - loss: 0.1193 - categorical_accuracy: 0.9638 - val_loss: 0.4650 - val_categorical_accuracy: 0.8400 - lr: 1.0000e-04\n",
      "Epoch 22/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1077 - categorical_accuracy: 0.9698\n",
      "Epoch 00022: val_loss did not improve from 0.36275\n",
      "67/67 [==============================] - 163s 2s/step - loss: 0.1077 - categorical_accuracy: 0.9698 - val_loss: 0.4709 - val_categorical_accuracy: 0.8400 - lr: 1.0000e-04\n",
      "Epoch 23/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1621 - categorical_accuracy: 0.9502\n",
      "Epoch 00023: val_loss did not improve from 0.36275\n",
      "67/67 [==============================] - 162s 2s/step - loss: 0.1621 - categorical_accuracy: 0.9502 - val_loss: 0.5492 - val_categorical_accuracy: 0.8400 - lr: 1.0000e-04\n",
      "Epoch 24/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0890 - categorical_accuracy: 0.9774\n",
      "Epoch 00024: val_loss did not improve from 0.36275\n",
      "67/67 [==============================] - 160s 2s/step - loss: 0.0890 - categorical_accuracy: 0.9774 - val_loss: 0.4296 - val_categorical_accuracy: 0.8600 - lr: 1.0000e-04\n",
      "Epoch 25/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1099 - categorical_accuracy: 0.9638\n",
      "Epoch 00025: val_loss did not improve from 0.36275\n",
      "67/67 [==============================] - 158s 2s/step - loss: 0.1099 - categorical_accuracy: 0.9638 - val_loss: 0.4900 - val_categorical_accuracy: 0.8300 - lr: 1.0000e-04\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(train_generator, steps_per_epoch=steps_per_epoch, epochs=num_epochs, verbose=1, \n",
    "                             callbacks=callbacks_list, validation_data=val_generator, \n",
    "                             validation_steps=validation_steps, class_weight=None, workers=1, initial_epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 590
    },
    "colab_type": "code",
    "id": "Yt2x8AOWCjEX",
    "outputId": "a23f6f71-2279-4a09-8b64-7645fe8d851e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['loss', 'categorical_accuracy', 'val_loss', 'val_categorical_accuracy', 'lr'])\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_loss_accuracy(history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "gLElBs3J0XGg"
   },
   "source": [
    "#### Model 2: Conv 3D Model with 20 frames per video (16, 32, 64, 128 filters conv 3D layers + 256 dense layer + 128 dense layer + image size 120 by 120)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "Na-JOEh-0XGm",
    "outputId": "b3fb2ba4-aa1f-4d17-9ee4-c9d2e05eed49"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# batch size = 10\n",
      "# num_frames per video = 20\n",
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv3d (Conv3D)             (None, 20, 120, 120, 16)  1312      \n",
      "                                                                 \n",
      " activation (Activation)     (None, 20, 120, 120, 16)  0         \n",
      "                                                                 \n",
      " batch_normalization (BatchN  (None, 20, 120, 120, 16)  64       \n",
      " ormalization)                                                   \n",
      "                                                                 \n",
      " max_pooling3d (MaxPooling3D  (None, 10, 60, 60, 16)   0         \n",
      " )                                                               \n",
      "                                                                 \n",
      " conv3d_1 (Conv3D)           (None, 10, 60, 60, 32)    13856     \n",
      "                                                                 \n",
      " activation_1 (Activation)   (None, 10, 60, 60, 32)    0         \n",
      "                                                                 \n",
      " batch_normalization_1 (Batc  (None, 10, 60, 60, 32)   128       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " max_pooling3d_1 (MaxPooling  (None, 5, 30, 30, 32)    0         \n",
      " 3D)                                                             \n",
      "                                                                 \n",
      " conv3d_2 (Conv3D)           (None, 5, 30, 30, 64)     55360     \n",
      "                                                                 \n",
      " activation_2 (Activation)   (None, 5, 30, 30, 64)     0         \n",
      "                                                                 \n",
      " batch_normalization_2 (Batc  (None, 5, 30, 30, 64)    256       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " max_pooling3d_2 (MaxPooling  (None, 2, 15, 15, 64)    0         \n",
      " 3D)                                                             \n",
      "                                                                 \n",
      " conv3d_3 (Conv3D)           (None, 2, 15, 15, 128)    221312    \n",
      "                                                                 \n",
      " activation_3 (Activation)   (None, 2, 15, 15, 128)    0         \n",
      "                                                                 \n",
      " batch_normalization_3 (Batc  (None, 2, 15, 15, 128)   512       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " max_pooling3d_3 (MaxPooling  (None, 1, 7, 7, 128)     0         \n",
      " 3D)                                                             \n",
      "                                                                 \n",
      " flatten (Flatten)           (None, 6272)              0         \n",
      "                                                                 \n",
      " dense (Dense)               (None, 256)               1605888   \n",
      "                                                                 \n",
      " activation_4 (Activation)   (None, 256)               0         \n",
      "                                                                 \n",
      " batch_normalization_4 (Batc  (None, 256)              1024      \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " dropout (Dropout)           (None, 256)               0         \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 128)               32896     \n",
      "                                                                 \n",
      " activation_5 (Activation)   (None, 128)               0         \n",
      "                                                                 \n",
      " batch_normalization_5 (Batc  (None, 128)              512       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " dropout_1 (Dropout)         (None, 128)               0         \n",
      "                                                                 \n",
      " dense_2 (Dense)             (None, 5)                 645       \n",
      "                                                                 \n",
      " activation_6 (Activation)   (None, 5)                 0         \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 1,933,765\n",
      "Trainable params: 1,932,517\n",
      "Non-trainable params: 1,248\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-08-29 13:41:37.456719: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:39] Overriding allow_growth setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.\n",
      "2023-08-29 13:41:37.456790: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1525] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 14796 MB memory:  -> device: 0, name: Quadro RTX 5000, pci bus id: 0000:40:00.0, compute capability: 7.5\n"
     ]
    }
   ],
   "source": [
    "num_epochs = 20 # choose the number of epochs\n",
    "#print ('# epochs =', num_epochs)\n",
    "batch_size = 10  #experiment with the batch size\n",
    "print ('# batch size =', batch_size)\n",
    "num_frames = 20\n",
    "print ('# num_frames per video =', num_frames)\n",
    "height = 120\n",
    "width = 120\n",
    "\n",
    "model = conv3D(num_images=num_frames)\n",
    "\n",
    "print (model.summary())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "4qHb4E6P0XGu"
   },
   "source": [
    "The `steps_per_epoch` and `validation_steps` are used by `fit_generator` to decide the number of next() calls it need to make.\n",
    "Let us now fit the model. This will start training the model and with the help of the checkpoints, you'll be able to save the model at the end of each epoch."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "EoVbyXK60XGv"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:`period` argument is deprecated. Please use `save_freq` to specify the frequency in number of batches seen.\n"
     ]
    }
   ],
   "source": [
    "callbacks_list = model_callbacks(\"model_conv3D_2\")\n",
    "\n",
    "steps_per_epoch, validation_steps = calculate_steps(num_train_sequences, num_val_sequences, batch_size)\n",
    "\n",
    "train_generator = generate(train_path, train_doc, batch_size, num_images=num_frames)\n",
    "\n",
    "val_generator   = generate(val_path, val_doc, batch_size, num_images=num_frames)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "uuVvXEwt0XGz",
    "outputId": "27661573-c2d3-41db-abeb-4a570f6054e9"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-08-29 13:41:57.212142: I tensorflow/stream_executor/cuda/cuda_dnn.cc:377] Loaded cuDNN version 8302\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "67/67 [==============================] - ETA: 0s - loss: 1.4720 - categorical_accuracy: 0.4661\n",
      "Epoch 00001: val_loss improved from inf to 6.74059, saving model to model_conv3D_2_2023-08-2913_41_03.339316/model-00001-1.47204-0.46606-6.74059-0.16000.h5\n",
      "67/67 [==============================] - 112s 2s/step - loss: 1.4720 - categorical_accuracy: 0.4661 - val_loss: 6.7406 - val_categorical_accuracy: 0.1600 - lr: 0.0010\n",
      "Epoch 2/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 1.0739 - categorical_accuracy: 0.5943\n",
      "Epoch 00002: val_loss did not improve from 6.74059\n",
      "67/67 [==============================] - 110s 2s/step - loss: 1.0739 - categorical_accuracy: 0.5943 - val_loss: 7.3428 - val_categorical_accuracy: 0.1500 - lr: 0.0010\n",
      "Epoch 3/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.7746 - categorical_accuracy: 0.7134\n",
      "Epoch 00003: val_loss improved from 6.74059 to 5.45189, saving model to model_conv3D_2_2023-08-2913_41_03.339316/model-00003-0.77455-0.71342-5.45189-0.24000.h5\n",
      "67/67 [==============================] - 109s 2s/step - loss: 0.7746 - categorical_accuracy: 0.7134 - val_loss: 5.4519 - val_categorical_accuracy: 0.2400 - lr: 0.0010\n",
      "Epoch 4/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.7455 - categorical_accuracy: 0.7104\n",
      "Epoch 00004: val_loss improved from 5.45189 to 5.05006, saving model to model_conv3D_2_2023-08-2913_41_03.339316/model-00004-0.74547-0.71041-5.05006-0.21000.h5\n",
      "67/67 [==============================] - 111s 2s/step - loss: 0.7455 - categorical_accuracy: 0.7104 - val_loss: 5.0501 - val_categorical_accuracy: 0.2100 - lr: 0.0010\n",
      "Epoch 5/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.6898 - categorical_accuracy: 0.7647\n",
      "Epoch 00005: val_loss improved from 5.05006 to 2.45139, saving model to model_conv3D_2_2023-08-2913_41_03.339316/model-00005-0.68979-0.76471-2.45139-0.34000.h5\n",
      "67/67 [==============================] - 109s 2s/step - loss: 0.6898 - categorical_accuracy: 0.7647 - val_loss: 2.4514 - val_categorical_accuracy: 0.3400 - lr: 0.0010\n",
      "Epoch 6/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.5761 - categorical_accuracy: 0.7888\n",
      "Epoch 00006: val_loss did not improve from 2.45139\n",
      "67/67 [==============================] - 110s 2s/step - loss: 0.5761 - categorical_accuracy: 0.7888 - val_loss: 3.4070 - val_categorical_accuracy: 0.2700 - lr: 0.0010\n",
      "Epoch 7/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.4732 - categorical_accuracy: 0.8250\n",
      "Epoch 00007: val_loss improved from 2.45139 to 2.22886, saving model to model_conv3D_2_2023-08-2913_41_03.339316/model-00007-0.47319-0.82504-2.22886-0.43000.h5\n",
      "67/67 [==============================] - 108s 2s/step - loss: 0.4732 - categorical_accuracy: 0.8250 - val_loss: 2.2289 - val_categorical_accuracy: 0.4300 - lr: 0.0010\n",
      "Epoch 8/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3749 - categorical_accuracy: 0.8627\n",
      "Epoch 00008: val_loss improved from 2.22886 to 1.86607, saving model to model_conv3D_2_2023-08-2913_41_03.339316/model-00008-0.37487-0.86275-1.86607-0.59000.h5\n",
      "67/67 [==============================] - 109s 2s/step - loss: 0.3749 - categorical_accuracy: 0.8627 - val_loss: 1.8661 - val_categorical_accuracy: 0.5900 - lr: 0.0010\n",
      "Epoch 9/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3438 - categorical_accuracy: 0.8718\n",
      "Epoch 00009: val_loss did not improve from 1.86607\n",
      "67/67 [==============================] - 109s 2s/step - loss: 0.3438 - categorical_accuracy: 0.8718 - val_loss: 3.4868 - val_categorical_accuracy: 0.4200 - lr: 0.0010\n",
      "Epoch 10/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3797 - categorical_accuracy: 0.8612\n",
      "Epoch 00010: val_loss improved from 1.86607 to 0.64132, saving model to model_conv3D_2_2023-08-2913_41_03.339316/model-00010-0.37967-0.86124-0.64132-0.80000.h5\n",
      "67/67 [==============================] - 110s 2s/step - loss: 0.3797 - categorical_accuracy: 0.8612 - val_loss: 0.6413 - val_categorical_accuracy: 0.8000 - lr: 0.0010\n",
      "Epoch 11/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2583 - categorical_accuracy: 0.9186\n",
      "Epoch 00011: val_loss did not improve from 0.64132\n",
      "67/67 [==============================] - 110s 2s/step - loss: 0.2583 - categorical_accuracy: 0.9186 - val_loss: 0.9704 - val_categorical_accuracy: 0.6900 - lr: 0.0010\n",
      "Epoch 12/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1862 - categorical_accuracy: 0.9321\n",
      "Epoch 00012: val_loss did not improve from 0.64132\n",
      "67/67 [==============================] - 109s 2s/step - loss: 0.1862 - categorical_accuracy: 0.9321 - val_loss: 4.5175 - val_categorical_accuracy: 0.4600 - lr: 0.0010\n",
      "Epoch 13/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2659 - categorical_accuracy: 0.9065\n",
      "Epoch 00013: val_loss did not improve from 0.64132\n",
      "\n",
      "Epoch 00013: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.\n",
      "67/67 [==============================] - 108s 2s/step - loss: 0.2659 - categorical_accuracy: 0.9065 - val_loss: 1.9374 - val_categorical_accuracy: 0.4600 - lr: 0.0010\n",
      "Epoch 14/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1809 - categorical_accuracy: 0.9367\n",
      "Epoch 00014: val_loss improved from 0.64132 to 0.43661, saving model to model_conv3D_2_2023-08-2913_41_03.339316/model-00014-0.18093-0.93665-0.43661-0.84000.h5\n",
      "67/67 [==============================] - 109s 2s/step - loss: 0.1809 - categorical_accuracy: 0.9367 - val_loss: 0.4366 - val_categorical_accuracy: 0.8400 - lr: 1.0000e-04\n",
      "Epoch 15/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1625 - categorical_accuracy: 0.9472\n",
      "Epoch 00015: val_loss did not improve from 0.43661\n",
      "67/67 [==============================] - 109s 2s/step - loss: 0.1625 - categorical_accuracy: 0.9472 - val_loss: 0.6711 - val_categorical_accuracy: 0.8100 - lr: 1.0000e-04\n",
      "Epoch 16/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1486 - categorical_accuracy: 0.9487\n",
      "Epoch 00016: val_loss improved from 0.43661 to 0.25382, saving model to model_conv3D_2_2023-08-2913_41_03.339316/model-00016-0.14857-0.94872-0.25382-0.87000.h5\n",
      "67/67 [==============================] - 110s 2s/step - loss: 0.1486 - categorical_accuracy: 0.9487 - val_loss: 0.2538 - val_categorical_accuracy: 0.8700 - lr: 1.0000e-04\n",
      "Epoch 17/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0926 - categorical_accuracy: 0.9698\n",
      "Epoch 00017: val_loss did not improve from 0.25382\n",
      "67/67 [==============================] - 108s 2s/step - loss: 0.0926 - categorical_accuracy: 0.9698 - val_loss: 0.2912 - val_categorical_accuracy: 0.8700 - lr: 1.0000e-04\n",
      "Epoch 18/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0833 - categorical_accuracy: 0.9759\n",
      "Epoch 00018: val_loss did not improve from 0.25382\n",
      "67/67 [==============================] - 107s 2s/step - loss: 0.0833 - categorical_accuracy: 0.9759 - val_loss: 0.3140 - val_categorical_accuracy: 0.8700 - lr: 1.0000e-04\n",
      "Epoch 19/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0750 - categorical_accuracy: 0.9804\n",
      "Epoch 00019: val_loss did not improve from 0.25382\n",
      "\n",
      "Epoch 00019: ReduceLROnPlateau reducing learning rate to 0.0001.\n",
      "67/67 [==============================] - 110s 2s/step - loss: 0.0750 - categorical_accuracy: 0.9804 - val_loss: 0.2629 - val_categorical_accuracy: 0.9100 - lr: 1.0000e-04\n",
      "Epoch 20/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1071 - categorical_accuracy: 0.9698\n",
      "Epoch 00020: val_loss improved from 0.25382 to 0.24425, saving model to model_conv3D_2_2023-08-2913_41_03.339316/model-00020-0.10707-0.96983-0.24425-0.90000.h5\n",
      "67/67 [==============================] - 109s 2s/step - loss: 0.1071 - categorical_accuracy: 0.9698 - val_loss: 0.2443 - val_categorical_accuracy: 0.9000 - lr: 1.0000e-04\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(train_generator, steps_per_epoch=steps_per_epoch, epochs=num_epochs, verbose=1, \n",
    "                              callbacks=callbacks_list, validation_data=val_generator, \n",
    "                              validation_steps=validation_steps, class_weight=None, workers=1, initial_epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 590
    },
    "colab_type": "code",
    "id": "7MYLaHHv0XG3",
    "outputId": "4e9ab984-c294-4bcd-bba5-0027fd8182bc"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['loss', 'categorical_accuracy', 'val_loss', 'val_categorical_accuracy', 'lr'])\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_loss_accuracy(history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "6M9yv49snDYY"
   },
   "source": [
    "The `steps_per_epoch` and `validation_steps` are used by `fit_generator` to decide the number of next() calls it need to make.\n",
    "Let us now fit the model. This will start training the model and with the help of the checkpoints, you'll be able to save the model at the end of each epoch."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "2Jm7_Pm0kcKC"
   },
   "source": [
    "#### Model 3: Conv 3D Model with 20 frames per video (16, 32, 64, 128 filters conv 3D layers + 256 dense layer + 128 dense layer + image size 160 by 160)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "3hhGIySIkcKF",
    "outputId": "4bd7d2be-f9ee-467e-cb41-1250631a85ed"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# epochs = 20\n",
      "# batch size = 10\n",
      "# num_frames per video = 20\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-08-29 16:28:46.609339: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:39] Overriding allow_growth setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.\n",
      "2023-08-29 16:28:46.609397: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1525] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 14796 MB memory:  -> device: 0, name: Quadro RTX 5000, pci bus id: 0000:1d:00.0, compute capability: 7.5\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv3d (Conv3D)             (None, 20, 160, 160, 16)  1312      \n",
      "                                                                 \n",
      " activation (Activation)     (None, 20, 160, 160, 16)  0         \n",
      "                                                                 \n",
      " batch_normalization (BatchN  (None, 20, 160, 160, 16)  64       \n",
      " ormalization)                                                   \n",
      "                                                                 \n",
      " max_pooling3d (MaxPooling3D  (None, 10, 80, 80, 16)   0         \n",
      " )                                                               \n",
      "                                                                 \n",
      " conv3d_1 (Conv3D)           (None, 10, 80, 80, 32)    13856     \n",
      "                                                                 \n",
      " activation_1 (Activation)   (None, 10, 80, 80, 32)    0         \n",
      "                                                                 \n",
      " batch_normalization_1 (Batc  (None, 10, 80, 80, 32)   128       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " max_pooling3d_1 (MaxPooling  (None, 5, 40, 40, 32)    0         \n",
      " 3D)                                                             \n",
      "                                                                 \n",
      " conv3d_2 (Conv3D)           (None, 5, 40, 40, 64)     55360     \n",
      "                                                                 \n",
      " activation_2 (Activation)   (None, 5, 40, 40, 64)     0         \n",
      "                                                                 \n",
      " batch_normalization_2 (Batc  (None, 5, 40, 40, 64)    256       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " max_pooling3d_2 (MaxPooling  (None, 2, 20, 20, 64)    0         \n",
      " 3D)                                                             \n",
      "                                                                 \n",
      " conv3d_3 (Conv3D)           (None, 2, 20, 20, 128)    221312    \n",
      "                                                                 \n",
      " activation_3 (Activation)   (None, 2, 20, 20, 128)    0         \n",
      "                                                                 \n",
      " batch_normalization_3 (Batc  (None, 2, 20, 20, 128)   512       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " max_pooling3d_3 (MaxPooling  (None, 1, 10, 10, 128)   0         \n",
      " 3D)                                                             \n",
      "                                                                 \n",
      " flatten (Flatten)           (None, 12800)             0         \n",
      "                                                                 \n",
      " dense (Dense)               (None, 256)               3277056   \n",
      "                                                                 \n",
      " activation_4 (Activation)   (None, 256)               0         \n",
      "                                                                 \n",
      " batch_normalization_4 (Batc  (None, 256)              1024      \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " dropout (Dropout)           (None, 256)               0         \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 128)               32896     \n",
      "                                                                 \n",
      " activation_5 (Activation)   (None, 128)               0         \n",
      "                                                                 \n",
      " batch_normalization_5 (Batc  (None, 128)              512       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " dropout_1 (Dropout)         (None, 128)               0         \n",
      "                                                                 \n",
      " dense_2 (Dense)             (None, 5)                 645       \n",
      "                                                                 \n",
      " activation_6 (Activation)   (None, 5)                 0         \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 3,604,933\n",
      "Trainable params: 3,603,685\n",
      "Non-trainable params: 1,248\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "num_epochs = 20 # choose the number of epochs\n",
    "print ('# epochs =', num_epochs)\n",
    "batch_size = 10  #experiment with the batch size\n",
    "print ('# batch size =', batch_size)\n",
    "num_frames = 20\n",
    "print ('# num_frames per video =', num_frames)\n",
    "height = 160\n",
    "width = 160\n",
    "\n",
    "model = conv3D(num_images=num_frames, height=160, width=160)\n",
    "\n",
    "print (model.summary())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "auyrgZ7FkcKH"
   },
   "source": [
    "The `steps_per_epoch` and `validation_steps` are used by `fit_generator` to decide the number of next() calls it need to make.\n",
    "Let us now fit the model. This will start training the model and with the help of the checkpoints, you'll be able to save the model at the end of each epoch."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "kQUViOEMkcKI"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:`period` argument is deprecated. Please use `save_freq` to specify the frequency in number of batches seen.\n"
     ]
    }
   ],
   "source": [
    "callbacks_list = model_callbacks(\"model_conv3D_3\")\n",
    "\n",
    "steps_per_epoch, validation_steps = calculate_steps(num_train_sequences, num_val_sequences, batch_size)\n",
    "\n",
    "train_generator = generate(train_path, train_doc, batch_size, num_images=num_frames, height=160, width=160)\n",
    "\n",
    "val_generator   = generate(val_path, val_doc, batch_size, num_images=num_frames, height=160, width=160)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "whHipbw3kcKJ",
    "outputId": "5eb2b4b4-8fc4-4a52-fbaf-92266b1f51c8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-08-29 16:29:00.608094: I tensorflow/stream_executor/cuda/cuda_dnn.cc:377] Loaded cuDNN version 8302\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "67/67 [==============================] - ETA: 0s - loss: 1.5180 - categorical_accuracy: 0.4434\n",
      "Epoch 00001: val_loss improved from inf to 6.51329, saving model to model_conv3D_3_2023-08-2916_28_22.343912/model-00001-1.51799-0.44344-6.51329-0.13000.h5\n",
      "67/67 [==============================] - 130s 2s/step - loss: 1.5180 - categorical_accuracy: 0.4434 - val_loss: 6.5133 - val_categorical_accuracy: 0.1300 - lr: 0.0010\n",
      "Epoch 2/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 1.0894 - categorical_accuracy: 0.5747\n",
      "Epoch 00002: val_loss did not improve from 6.51329\n",
      "67/67 [==============================] - 123s 2s/step - loss: 1.0894 - categorical_accuracy: 0.5747 - val_loss: 8.1363 - val_categorical_accuracy: 0.2100 - lr: 0.0010\n",
      "Epoch 3/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.8435 - categorical_accuracy: 0.6802\n",
      "Epoch 00003: val_loss did not improve from 6.51329\n",
      "67/67 [==============================] - 122s 2s/step - loss: 0.8435 - categorical_accuracy: 0.6802 - val_loss: 7.0792 - val_categorical_accuracy: 0.2400 - lr: 0.0010\n",
      "Epoch 4/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.6368 - categorical_accuracy: 0.7753\n",
      "Epoch 00004: val_loss did not improve from 6.51329\n",
      "\n",
      "Epoch 00004: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.\n",
      "67/67 [==============================] - 123s 2s/step - loss: 0.6368 - categorical_accuracy: 0.7753 - val_loss: 7.6413 - val_categorical_accuracy: 0.2200 - lr: 0.0010\n",
      "Epoch 5/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.4735 - categorical_accuracy: 0.8145\n",
      "Epoch 00005: val_loss improved from 6.51329 to 5.16376, saving model to model_conv3D_3_2023-08-2916_28_22.343912/model-00005-0.47353-0.81448-5.16376-0.31000.h5\n",
      "67/67 [==============================] - 121s 2s/step - loss: 0.4735 - categorical_accuracy: 0.8145 - val_loss: 5.1638 - val_categorical_accuracy: 0.3100 - lr: 1.0000e-04\n",
      "Epoch 6/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.4193 - categorical_accuracy: 0.8477\n",
      "Epoch 00006: val_loss improved from 5.16376 to 4.05275, saving model to model_conv3D_3_2023-08-2916_28_22.343912/model-00006-0.41934-0.84766-4.05275-0.25000.h5\n",
      "67/67 [==============================] - 120s 2s/step - loss: 0.4193 - categorical_accuracy: 0.8477 - val_loss: 4.0528 - val_categorical_accuracy: 0.2500 - lr: 1.0000e-04\n",
      "Epoch 7/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.4040 - categorical_accuracy: 0.8552\n",
      "Epoch 00007: val_loss improved from 4.05275 to 2.76193, saving model to model_conv3D_3_2023-08-2916_28_22.343912/model-00007-0.40405-0.85520-2.76193-0.42000.h5\n",
      "67/67 [==============================] - 116s 2s/step - loss: 0.4040 - categorical_accuracy: 0.8552 - val_loss: 2.7619 - val_categorical_accuracy: 0.4200 - lr: 1.0000e-04\n",
      "Epoch 8/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3337 - categorical_accuracy: 0.8748\n",
      "Epoch 00008: val_loss improved from 2.76193 to 1.84233, saving model to model_conv3D_3_2023-08-2916_28_22.343912/model-00008-0.33375-0.87481-1.84233-0.46000.h5\n",
      "67/67 [==============================] - 118s 2s/step - loss: 0.3337 - categorical_accuracy: 0.8748 - val_loss: 1.8423 - val_categorical_accuracy: 0.4600 - lr: 1.0000e-04\n",
      "Epoch 9/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3459 - categorical_accuracy: 0.8778\n",
      "Epoch 00009: val_loss improved from 1.84233 to 1.24713, saving model to model_conv3D_3_2023-08-2916_28_22.343912/model-00009-0.34594-0.87783-1.24713-0.60000.h5\n",
      "67/67 [==============================] - 120s 2s/step - loss: 0.3459 - categorical_accuracy: 0.8778 - val_loss: 1.2471 - val_categorical_accuracy: 0.6000 - lr: 1.0000e-04\n",
      "Epoch 10/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2830 - categorical_accuracy: 0.9035\n",
      "Epoch 00010: val_loss improved from 1.24713 to 0.63503, saving model to model_conv3D_3_2023-08-2916_28_22.343912/model-00010-0.28302-0.90347-0.63503-0.74000.h5\n",
      "67/67 [==============================] - 118s 2s/step - loss: 0.2830 - categorical_accuracy: 0.9035 - val_loss: 0.6350 - val_categorical_accuracy: 0.7400 - lr: 1.0000e-04\n",
      "Epoch 11/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2836 - categorical_accuracy: 0.9050\n",
      "Epoch 00011: val_loss did not improve from 0.63503\n",
      "67/67 [==============================] - 117s 2s/step - loss: 0.2836 - categorical_accuracy: 0.9050 - val_loss: 0.8630 - val_categorical_accuracy: 0.7300 - lr: 1.0000e-04\n",
      "Epoch 12/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2518 - categorical_accuracy: 0.9201\n",
      "Epoch 00012: val_loss did not improve from 0.63503\n",
      "67/67 [==============================] - 122s 2s/step - loss: 0.2518 - categorical_accuracy: 0.9201 - val_loss: 1.0109 - val_categorical_accuracy: 0.7000 - lr: 1.0000e-04\n",
      "Epoch 13/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2691 - categorical_accuracy: 0.9050\n",
      "Epoch 00013: val_loss improved from 0.63503 to 0.33106, saving model to model_conv3D_3_2023-08-2916_28_22.343912/model-00013-0.26910-0.90498-0.33106-0.89000.h5\n",
      "67/67 [==============================] - 119s 2s/step - loss: 0.2691 - categorical_accuracy: 0.9050 - val_loss: 0.3311 - val_categorical_accuracy: 0.8900 - lr: 1.0000e-04\n",
      "Epoch 14/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2176 - categorical_accuracy: 0.9201\n",
      "Epoch 00014: val_loss improved from 0.33106 to 0.30410, saving model to model_conv3D_3_2023-08-2916_28_22.343912/model-00014-0.21762-0.92006-0.30410-0.90000.h5\n",
      "67/67 [==============================] - 116s 2s/step - loss: 0.2176 - categorical_accuracy: 0.9201 - val_loss: 0.3041 - val_categorical_accuracy: 0.9000 - lr: 1.0000e-04\n",
      "Epoch 15/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2262 - categorical_accuracy: 0.9201\n",
      "Epoch 00015: val_loss did not improve from 0.30410\n",
      "67/67 [==============================] - 116s 2s/step - loss: 0.2262 - categorical_accuracy: 0.9201 - val_loss: 0.4492 - val_categorical_accuracy: 0.8400 - lr: 1.0000e-04\n",
      "Epoch 16/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1884 - categorical_accuracy: 0.9412\n",
      "Epoch 00016: val_loss did not improve from 0.30410\n",
      "67/67 [==============================] - 118s 2s/step - loss: 0.1884 - categorical_accuracy: 0.9412 - val_loss: 0.5976 - val_categorical_accuracy: 0.7700 - lr: 1.0000e-04\n",
      "Epoch 17/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1783 - categorical_accuracy: 0.9397\n",
      "Epoch 00017: val_loss did not improve from 0.30410\n",
      "\n",
      "Epoch 00017: ReduceLROnPlateau reducing learning rate to 0.0001.\n",
      "67/67 [==============================] - 118s 2s/step - loss: 0.1783 - categorical_accuracy: 0.9397 - val_loss: 0.3474 - val_categorical_accuracy: 0.8700 - lr: 1.0000e-04\n",
      "Epoch 18/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1395 - categorical_accuracy: 0.9517\n",
      "Epoch 00018: val_loss did not improve from 0.30410\n",
      "67/67 [==============================] - 115s 2s/step - loss: 0.1395 - categorical_accuracy: 0.9517 - val_loss: 0.3193 - val_categorical_accuracy: 0.8700 - lr: 1.0000e-04\n",
      "Epoch 19/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1170 - categorical_accuracy: 0.9698\n",
      "Epoch 00019: val_loss did not improve from 0.30410\n",
      "67/67 [==============================] - 116s 2s/step - loss: 0.1170 - categorical_accuracy: 0.9698 - val_loss: 0.3071 - val_categorical_accuracy: 0.8800 - lr: 1.0000e-04\n",
      "Epoch 20/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1966 - categorical_accuracy: 0.9321\n",
      "Epoch 00020: val_loss improved from 0.30410 to 0.28305, saving model to model_conv3D_3_2023-08-2916_28_22.343912/model-00020-0.19659-0.93213-0.28305-0.91000.h5\n",
      "67/67 [==============================] - 119s 2s/step - loss: 0.1966 - categorical_accuracy: 0.9321 - val_loss: 0.2831 - val_categorical_accuracy: 0.9100 - lr: 1.0000e-04\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(train_generator, steps_per_epoch=steps_per_epoch, epochs=num_epochs, verbose=1, \n",
    "                              callbacks=callbacks_list, validation_data=val_generator, \n",
    "                              validation_steps=validation_steps, class_weight=None, workers=1, initial_epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 590
    },
    "colab_type": "code",
    "id": "628-8CImkcKM",
    "outputId": "95790ba5-73c9-46bd-9a99-24494d7b191e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['loss', 'categorical_accuracy', 'val_loss', 'val_categorical_accuracy', 'lr'])\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_loss_accuracy(history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "aA5W8giixDpY"
   },
   "source": [
    "#### Conv 3D Model with 30 frames per video (16, 32, 64, 128 filters conv 3D layers + 256 dense layer + 128 dense layer) + Random data transformations on training data set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "hWDBO-jK1UHu",
    "outputId": "44333e08-11db-494d-93c7-7978150db952"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# epochs = 15\n",
      "# batch size = 10\n",
      "# num_frames per video = 30\n",
      "Model: \"sequential_4\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv3d_16 (Conv3D)          (None, 30, 120, 120, 16)  1312      \n",
      "                                                                 \n",
      " activation_28 (Activation)  (None, 30, 120, 120, 16)  0         \n",
      "                                                                 \n",
      " batch_normalization_24 (Bat  (None, 30, 120, 120, 16)  64       \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " max_pooling3d_16 (MaxPoolin  (None, 15, 60, 60, 16)   0         \n",
      " g3D)                                                            \n",
      "                                                                 \n",
      " conv3d_17 (Conv3D)          (None, 15, 60, 60, 32)    13856     \n",
      "                                                                 \n",
      " activation_29 (Activation)  (None, 15, 60, 60, 32)    0         \n",
      "                                                                 \n",
      " batch_normalization_25 (Bat  (None, 15, 60, 60, 32)   128       \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " max_pooling3d_17 (MaxPoolin  (None, 7, 30, 30, 32)    0         \n",
      " g3D)                                                            \n",
      "                                                                 \n",
      " conv3d_18 (Conv3D)          (None, 7, 30, 30, 64)     55360     \n",
      "                                                                 \n",
      " activation_30 (Activation)  (None, 7, 30, 30, 64)     0         \n",
      "                                                                 \n",
      " batch_normalization_26 (Bat  (None, 7, 30, 30, 64)    256       \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " max_pooling3d_18 (MaxPoolin  (None, 3, 15, 15, 64)    0         \n",
      " g3D)                                                            \n",
      "                                                                 \n",
      " conv3d_19 (Conv3D)          (None, 3, 15, 15, 128)    221312    \n",
      "                                                                 \n",
      " activation_31 (Activation)  (None, 3, 15, 15, 128)    0         \n",
      "                                                                 \n",
      " batch_normalization_27 (Bat  (None, 3, 15, 15, 128)   512       \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " max_pooling3d_19 (MaxPoolin  (None, 1, 7, 7, 128)     0         \n",
      " g3D)                                                            \n",
      "                                                                 \n",
      " flatten_4 (Flatten)         (None, 6272)              0         \n",
      "                                                                 \n",
      " dense_12 (Dense)            (None, 256)               1605888   \n",
      "                                                                 \n",
      " activation_32 (Activation)  (None, 256)               0         \n",
      "                                                                 \n",
      " batch_normalization_28 (Bat  (None, 256)              1024      \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " dropout_8 (Dropout)         (None, 256)               0         \n",
      "                                                                 \n",
      " dense_13 (Dense)            (None, 128)               32896     \n",
      "                                                                 \n",
      " activation_33 (Activation)  (None, 128)               0         \n",
      "                                                                 \n",
      " batch_normalization_29 (Bat  (None, 128)              512       \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " dropout_9 (Dropout)         (None, 128)               0         \n",
      "                                                                 \n",
      " dense_14 (Dense)            (None, 5)                 645       \n",
      "                                                                 \n",
      " activation_34 (Activation)  (None, 5)                 0         \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 1,933,765\n",
      "Trainable params: 1,932,517\n",
      "Non-trainable params: 1,248\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "num_epochs = 15 # choose the number of epochs\n",
    "print ('# epochs =', num_epochs)\n",
    "batch_size = 10  #experiment with the batch size\n",
    "print ('# batch size =', batch_size)\n",
    "num_frames = 30\n",
    "print ('# num_frames per video =', num_frames)\n",
    "height = 120\n",
    "width = 120\n",
    "\n",
    "model = conv3D(num_images=num_frames)\n",
    "\n",
    "print (model.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "VrPewKo_w9B-"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:`period` argument is deprecated. Please use `save_freq` to specify the frequency in number of batches seen.\n"
     ]
    }
   ],
   "source": [
    "callbacks_list = model_callbacks(\"model_conv3D_4\")\n",
    "\n",
    "steps_per_epoch, validation_steps = calculate_steps(num_train_sequences, num_val_sequences, batch_size)\n",
    "\n",
    "train_generator = generate(train_path, train_doc, batch_size, augment=True, num_images=num_frames)\n",
    "\n",
    "val_generator   = generate(val_path, val_doc, batch_size, num_images=num_frames)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "zNrzgwU9w9CE",
    "outputId": "e0222594-e5f6-414c-a240-30752a2e1a57"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 1.6539 - categorical_accuracy: 0.4103\n",
      "Epoch 00001: val_loss improved from inf to 1.65786, saving model to model_conv3D_5_2023-08-2812_46_28.519095/model-00001-1.65395-0.41026-1.65786-0.26000.h5\n",
      "67/67 [==============================] - 187s 3s/step - loss: 1.6539 - categorical_accuracy: 0.4103 - val_loss: 1.6579 - val_categorical_accuracy: 0.2600 - lr: 0.0010\n",
      "Epoch 2/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 1.2791 - categorical_accuracy: 0.5249\n",
      "Epoch 00002: val_loss did not improve from 1.65786\n",
      "67/67 [==============================] - 183s 3s/step - loss: 1.2791 - categorical_accuracy: 0.5249 - val_loss: 4.8654 - val_categorical_accuracy: 0.1700 - lr: 0.0010\n",
      "Epoch 3/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 1.1433 - categorical_accuracy: 0.5566\n",
      "Epoch 00003: val_loss did not improve from 1.65786\n",
      "67/67 [==============================] - 183s 3s/step - loss: 1.1433 - categorical_accuracy: 0.5566 - val_loss: 6.0084 - val_categorical_accuracy: 0.1700 - lr: 0.0010\n",
      "Epoch 4/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 1.0701 - categorical_accuracy: 0.6078\n",
      "Epoch 00004: val_loss did not improve from 1.65786\n",
      "\n",
      "Epoch 00004: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.\n",
      "67/67 [==============================] - 184s 3s/step - loss: 1.0701 - categorical_accuracy: 0.6078 - val_loss: 5.2923 - val_categorical_accuracy: 0.1400 - lr: 0.0010\n",
      "Epoch 5/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.8212 - categorical_accuracy: 0.6606\n",
      "Epoch 00005: val_loss did not improve from 1.65786\n",
      "67/67 [==============================] - 185s 3s/step - loss: 0.8212 - categorical_accuracy: 0.6606 - val_loss: 3.3264 - val_categorical_accuracy: 0.3100 - lr: 1.0000e-04\n",
      "Epoch 6/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.7717 - categorical_accuracy: 0.7134\n",
      "Epoch 00006: val_loss did not improve from 1.65786\n",
      "67/67 [==============================] - 185s 3s/step - loss: 0.7717 - categorical_accuracy: 0.7134 - val_loss: 2.1110 - val_categorical_accuracy: 0.4100 - lr: 1.0000e-04\n",
      "Epoch 7/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.7432 - categorical_accuracy: 0.7164\n",
      "Epoch 00007: val_loss did not improve from 1.65786\n",
      "\n",
      "Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.0001.\n",
      "67/67 [==============================] - 187s 3s/step - loss: 0.7432 - categorical_accuracy: 0.7164 - val_loss: 2.0285 - val_categorical_accuracy: 0.3200 - lr: 1.0000e-04\n",
      "Epoch 8/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.7126 - categorical_accuracy: 0.7330\n",
      "Epoch 00008: val_loss improved from 1.65786 to 1.52294, saving model to model_conv3D_5_2023-08-2812_46_28.519095/model-00008-0.71259-0.73303-1.52294-0.56000.h5\n",
      "67/67 [==============================] - 185s 3s/step - loss: 0.7126 - categorical_accuracy: 0.7330 - val_loss: 1.5229 - val_categorical_accuracy: 0.5600 - lr: 1.0000e-04\n",
      "Epoch 9/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.6854 - categorical_accuracy: 0.7149\n",
      "Epoch 00009: val_loss improved from 1.52294 to 1.36922, saving model to model_conv3D_5_2023-08-2812_46_28.519095/model-00009-0.68536-0.71493-1.36922-0.57000.h5\n",
      "67/67 [==============================] - 187s 3s/step - loss: 0.6854 - categorical_accuracy: 0.7149 - val_loss: 1.3692 - val_categorical_accuracy: 0.5700 - lr: 1.0000e-04\n",
      "Epoch 10/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.6200 - categorical_accuracy: 0.7858\n",
      "Epoch 00010: val_loss improved from 1.36922 to 1.07925, saving model to model_conv3D_5_2023-08-2812_46_28.519095/model-00010-0.62002-0.78582-1.07925-0.62000.h5\n",
      "67/67 [==============================] - 185s 3s/step - loss: 0.6200 - categorical_accuracy: 0.7858 - val_loss: 1.0792 - val_categorical_accuracy: 0.6200 - lr: 1.0000e-04\n",
      "Epoch 11/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.5953 - categorical_accuracy: 0.7783\n",
      "Epoch 00011: val_loss did not improve from 1.07925\n",
      "67/67 [==============================] - 185s 3s/step - loss: 0.5953 - categorical_accuracy: 0.7783 - val_loss: 1.3642 - val_categorical_accuracy: 0.5100 - lr: 1.0000e-04\n",
      "Epoch 12/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.6101 - categorical_accuracy: 0.7557\n",
      "Epoch 00012: val_loss improved from 1.07925 to 1.03167, saving model to model_conv3D_5_2023-08-2812_46_28.519095/model-00012-0.61006-0.75566-1.03167-0.62000.h5\n",
      "67/67 [==============================] - 187s 3s/step - loss: 0.6101 - categorical_accuracy: 0.7557 - val_loss: 1.0317 - val_categorical_accuracy: 0.6200 - lr: 1.0000e-04\n",
      "Epoch 13/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.5318 - categorical_accuracy: 0.7858\n",
      "Epoch 00013: val_loss improved from 1.03167 to 0.81916, saving model to model_conv3D_5_2023-08-2812_46_28.519095/model-00013-0.53175-0.78582-0.81916-0.65000.h5\n",
      "67/67 [==============================] - 186s 3s/step - loss: 0.5318 - categorical_accuracy: 0.7858 - val_loss: 0.8192 - val_categorical_accuracy: 0.6500 - lr: 1.0000e-04\n",
      "Epoch 14/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.5341 - categorical_accuracy: 0.7994\n",
      "Epoch 00014: val_loss improved from 0.81916 to 0.75588, saving model to model_conv3D_5_2023-08-2812_46_28.519095/model-00014-0.53411-0.79940-0.75588-0.74000.h5\n",
      "67/67 [==============================] - 184s 3s/step - loss: 0.5341 - categorical_accuracy: 0.7994 - val_loss: 0.7559 - val_categorical_accuracy: 0.7400 - lr: 1.0000e-04\n",
      "Epoch 15/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.5229 - categorical_accuracy: 0.7919\n",
      "Epoch 00015: val_loss did not improve from 0.75588\n",
      "67/67 [==============================] - 185s 3s/step - loss: 0.5229 - categorical_accuracy: 0.7919 - val_loss: 1.0281 - val_categorical_accuracy: 0.5300 - lr: 1.0000e-04\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(train_generator, steps_per_epoch=steps_per_epoch, epochs=num_epochs, verbose=1, \n",
    "                              callbacks=callbacks_list, validation_data=val_generator, \n",
    "                              validation_steps=validation_steps, class_weight=None, workers=1, initial_epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "_VZoNY-8w9CG"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['loss', 'categorical_accuracy', 'val_loss', 'val_categorical_accuracy', 'lr'])\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8/fFQqAAAACXBIWXMAAAsTAAALEwEAmpwYAABBFElEQVR4nO3dd3xUVfrH8c+TTgkdQu+9CRIQxIaCgihFLCDYXdx1VVZd69pWV1d/uq6ra8NeQUXaKk0sYKEXpQZCDzX0mpDy/P44AwZIYICZ3CnP+/XKK1Pu3PtNIPPMPefcc0RVMcYYE71ivA5gjDHGW1YIjDEmylkhMMaYKGeFwBhjopwVAmOMiXJWCIwxJspZITBRRUTeF5F/+LntahHpGuxMxnjNCoExxkQ5KwTGhCERifM6g4kcVghMyPE1ydwnIr+JyD4ReUdEUkRkvIjsEZHJIlK+wPa9RGSRiOwUkR9EpFmB59qKyFzf6z4Dko461mUiMt/32l9EpLWfGXuKyDwR2S0i60TkiaOeP8e3v52+52/0PV5CRP4lImtEZJeI/OR77AIRySjk99DVd/sJERkhIh+LyG7gRhHpICLTfMfYKCL/FZGEAq9vISLfiMh2EdksIg+LSFUR2S8iFQtsd6aIZIpIvD8/u4k8VghMqOoHdAMaA5cD44GHgcq4/7d3AYhIY2AY8Bffc+OA/4lIgu9NcTTwEVAB+MK3X3yvbQu8C9wGVATeBMaKSKIf+fYB1wPlgJ7An0Skj2+/dXx5X/FlagPM973uBaAdcLYv0/1Avp+/k97ACN8xPwHygLuBSkAn4CLgdl+GZGAyMAGoDjQEvlXVTcAPwNUF9nsdMFxVc/zMYSKMFQITql5R1c2quh74EZihqvNUNQsYBbT1bXcN8LWqfuN7I3sBKIF7o+0IxAMvqWqOqo4AZhU4xmDgTVWdoap5qvoBkO173XGp6g+qukBV81X1N1wxOt/39LXAZFUd5jvuNlWdLyIxwM3AEFVd7zvmL6qa7efvZJqqjvYd84CqzlHV6aqaq6qrcYXsUIbLgE2q+i9VzVLVPao6w/fcB8AgABGJBQbgiqWJUlYITKjaXOD2gULul/bdrg6sOfSEquYD64AavufW65EzK64pcLsOcK+vaWWniOwEavled1wicpaIfO9rUtkF/BH3yRzfPlYU8rJKuKapwp7zx7qjMjQWka9EZJOvuegZPzIAjAGai0g93FnXLlWdeYqZTASwQmDC3QbcGzoAIiK4N8H1wEaghu+xQ2oXuL0OeFpVyxX4Kqmqw/w47qfAWKCWqpYF3gAOHWcd0KCQ12wFsop4bh9QssDPEYtrViro6KmCXweWAo1UtQyu6axghvqFBfedVX2OOyu4DjsbiHpWCEy4+xzoKSIX+To778U17/wCTANygbtEJF5ErgA6FHjtW8AffZ/uRURK+TqBk/04bjKwXVWzRKQDrjnokE+AriJytYjEiUhFEWnjO1t5F3hRRKqLSKyIdPL1SSwDknzHjwceAU7UV5EM7Ab2ikhT4E8FnvsKqCYifxGRRBFJFpGzCjz/IXAj0AsrBFHPCoEJa6qahvtk+wruE/flwOWqelBVDwJX4N7wtuP6E0YWeO1s4A/Af4EdQLpvW3/cDjwpInuAx3AF6dB+1wKX4orSdlxH8Rm+p/8KLMD1VWwHngNiVHWXb59v485m9gFHjCIqxF9xBWgPrqh9ViDDHlyzz+XAJmA50KXA8z/jOqnnqmrB5jIThcQWpjEmOonId8Cnqvq211mMt6wQGBOFRKQ98A2uj2OP13mMt6xpyJgoIyIf4K4x+IsVAQN2RmCMMVHPzgiMMSbKhd3EVZUqVdK6det6HcMYY8LKnDlztqrq0demAGFYCOrWrcvs2bO9jmGMMWFFRIocJmxNQ8YYE+WsEBhjTJSzQmCMMVEu7PoICpOTk0NGRgZZWVleRwmqpKQkatasSXy8rR9ijAmciCgEGRkZJCcnU7duXY6caDJyqCrbtm0jIyODevXqeR3HGBNBgto0JCLdRSRNRNJF5MFCnq/tm9N9nrhlCS89leNkZWVRsWLFiC0CACJCxYoVI/6sxxhT/IJWCHzzqb8K9ACaAwNEpPlRmz0CfK6qbYH+wGuncbxTfWnYiIaf0RhT/ILZNNQBSFfVlQAiMhy35uriAtsoUMZ3uyxukRFjjAk7O/cfZPHG3SzZuIe8/HxSyiSRUiaJqr7vJRJivY5YpGAWghocubReBnDWUds8AUwSkTuBUkDXwnYkIoNx68tSu3btwjbx1M6dO/n000+5/fbbT+p1l156KZ9++inlypULTjBjTMDl5ysZOw6weOMuFm/YzeKNu1m8YTcbdh2/2bZMUhxVyyYdLhApZRIPF4mUMklULZtEpdKJxMYU/5m/153FA4D3VfVfItIJ+EhEWvpWcjpMVYcCQwFSU1NDbpa8nTt38tprrx1TCHJzc4mLK/pXPG7cuGBHM8achqycPJZv3nv4TX/Jxj0s2bibPdm5AMQINKhcmvb1KtC8WhmaVy9Ds2plSIyLYfPuLDbtynbfd2ex2fe1aXc2yzdvJXNvNnn5R76dxQhUTj62QFRJTqRq2SSaVi1D5eQTLVx38oJZCNbj1o49pKbvsYJuAboDqOo0EUnCLb69JYi5Au7BBx9kxYoVtGnThvj4eJKSkihfvjxLly5l2bJl9OnTh3Xr1pGVlcWQIUMYPHgw8Pt0GXv37qVHjx6cc845/PLLL9SoUYMxY8ZQokQJj38yY6LH9n0HWeL7dH/oU3565t7Db9alEmJpVq0Mfc+sQfNq7g2/SdVkkuILb/JJToqnYZWiVz3Ny1e27c32FQn3fcvuLDbtcoVjzbb9zFi1nV0Hcg6/5qk+LbmuY50i93mqglkIZgGNRKQergD058h1XQHWAhcB74tIMyAJyDydg/79f4tYvGH36eziGM2rl+Hxy1sU+fyzzz7LwoULmT9/Pj/88AM9e/Zk4cKFh4d5vvvuu1SoUIEDBw7Qvn17+vXrR8WKFY/Yx/Llyxk2bBhvvfUWV199NV9++SWDBg0K6M9hzCHz1+0kbdNuqvjasKuWSaJcyfioGZCgqsxbt5Pvlmw5/Ka/affvTTtVyyTRvHoZujVPoXn1MjSvVobaFUoSE8Bmm9gYoUqZJKqUSTrudgcO5rFljysQtSuWDNjxCwpaIVDVXBG5A5gIxALvquoiEXkSmK2qY3Frur4lInfjOo5v1AhYIKFDhw5HjPV/+eWXGTVqFADr1q1j+fLlxxSCevXq0aZNGwDatWvH6tWriyuuiTJf/baBIcPnH9MskRAXQ0qZRFKSk0gpe6iTM/GIDs+qZZOK/AQcDg4czGPM/PV8NH0NizbsJjZGaFi5NJ0aVDz8Kb9ZtWQqlg5888upKpEQS52KpahTsVTQjhHUPgJVHQeMO+qxxwrcXgx0DuQxj/fJvbiUKvX7P9gPP/zA5MmTmTZtGiVLluSCCy4o9FqAxMTf/+PFxsZy4MCBYslqosvoeeu55/P5tKtTnmf7tWbn/oOuWWJXwTbsLBZv2M13S7ZwICfvmH0c3elZtYwrHCnJiVQrW4ImVZNJiAut2WtWZu7l4+lrGTFnHbuzcmlaNZl/9GlJn7Y1KJ3odVep9+w3EADJycns2VP4in+7du2ifPnylCxZkqVLlzJ9+vRiTmeMM2JOBveN+JWO9Sryzo2plEw4/p+/qrInO5fNu35vwz5cLHZlsXlP4Z2eyYlxdGlahYtbpHB+48okJ3kzJUpuXj7fLd3CR9PX8OPyrcTHCt1bVuP6TnVIrVM+aprB/GGFIAAqVqxI586dadmyJSVKlCAlJeXwc927d+eNN96gWbNmNGnShI4dO3qY1ESr4TPX8tCoBZzTsBJDr0v1a0y7iFAmKZ4ySfE0SvGv03Pd9gNMXZbJ5CWbGfvrBhJiY+jUoCIXt0ihW7OUE7aHB0Lmnmw+n72OT6avYcOuLKqVTeLebo25pkMtqiQH//jhKOzWLE5NTdWjF6ZZsmQJzZo18yhR8Yqmn9UExkfT1/Do6IVc0KQybwxqVyxt/Hn5yty1O5i0aBOTFm9mzbb9ALStXY6Lm1fl4hYpNKhcOmDHU1XmrNnBh9PWMH7hRnLylHMaVmJQxzp0bVaFuNjQaqrygojMUdXUwp6zMwJjIti7P63iya8W07VZFV4deCaJccXT0RsbI7SvW4H2dSvw8KXNWL5l7+Gi8NyEpTw3YSkNKpeim68otKlZ7pRG5OzLzmXM/A18NH0NSzbuJjkpjkEd6zCoY52AFppIZ4XAmAg1dOoKnhm3lO4tqvLygLaedeCKCI1TkmmckswdFzZiw84DTF6ymUmLNvP2jyt5Y8oKKicn0q15Chc3T6FTg4onLFjpW/by8fQ1fDkngz3ZuTSrVoZ/XtGK3m2qn7DvwxzLfmPGRKBXv0/n+YlpXNa6Gv++pg3xIdQ0Ur1cCa7vVJfrO9Vl14EcfkjbwqRFmxkzbz2fzlhL6cQ4zm9SmYubp9ClaRXK+Dqbc/PymbxkMx9NX8PP6dtIiI3h0lZVua5THc6sbZ2/p8MKgTERRFV5afJy/vPtcvq2rcHzV7YO6fbxsiXi6d2mBr3b1CArJ49pK7YxafEmvlm8ha9/20h8rNCxfkWaVSvD2Pkb2LQ7ixrlSnDfJU24pn0tKoXQeP9wZoXAmAihqrwwKY1Xv1/BVe1q8my/1p5MYHaqkuJj6dK0Cl2aVuHpPu7K30mLNzFp0WZ+XL6VcxtV4qk+LbmwaZWw+rnCgRUCYyKAqvLs+KW8OXUlAzrU5uk+LQM6HUJxi4kR2tUpT7s65Xmwe1MO5ORZ238Qhe45Yxg5NPvoqXjppZfYv39/gBOZaKKqPPnVYt6cupLrO9Xhmb7hXQSOJiJWBILMCkEAWCEwXsnPVx4ds5D3fl7NzZ3r8fdeLazT1Jw0K7MBUHAa6m7dulGlShU+//xzsrOz6du3L3//+9/Zt28fV199NRkZGeTl5fHoo4+yefNmNmzYQJcuXahUqRLff/+91z+KCSP5+crDoxYwfNY6bju/Pg92b2pFwGuqMP01qNAAmnT3Oo3fIq8QjH8QNi0I7D6rtoIezxb5dMFpqCdNmsSIESOYOXMmqkqvXr2YOnUqmZmZVK9ena+//hpwcxCVLVuWF198ke+//55KlSoFNrOJaHn5yv0jfuPLuRnceWFD7unW2IpAKJj1Nkx8GMrXhcaXQJj8m0ReIfDYpEmTmDRpEm3btgVg7969LF++nHPPPZd7772XBx54gMsuu4xzzz3X46QmEPLylbRNe5izdgcrtuylebUydKxfkVoVSgTtjTk3L597v/iVMfM3cE+3xtx1UaOgHMecpNU/w4QHoXRV2LEa1s+Fmu28TuWXyCsEx/nkXhxUlYceeojbbrvtmOfmzp3LuHHjeOSRR7jooot47LHHCtmDCWV7s3OZt3YHc9a4r3lrd7LXt2xhQlwMB3PdKqvVyybRsX7Fw1+BKgw5efn8Zfh8vl6wkfu7N+H2Cxqe9j5NAOzKgM+vh/L14LqR8Eo7WPilFYJoUnAa6ksuuYRHH32UgQMHUrp0adavX098fDy5ublUqFCBQYMGUa5cOd5+++0jXmtNQ6FH1S1SPnftDmavdm/8SzftJl/dGX+TlGR6t6lOat3ypNapQI1yJViRuZfpK7cxfeV2pizLZOQ8tzprIArDwdx87hw2l4mLNvNIz2bcem79YPzY5mTlHIDh10LeQRgwDMrVhobdYNFIuPgfEBP6Y3KsEARAwWmoe/TowbXXXkunTp0AKF26NB9//DHp6encd999xMTEEB8fz+uvvw7A4MGD6d69O9WrV7fOYo/l5OWzeMNuZq/Zwdw1O5i9Zjubd2cDbr3atrXLc8eFjUitU542tcsdnvqgoEYpyTRKSea6TnVRVdK3FF4Yqh0uDBXoWL8itSuUPG5hyM7N4/aP5/Lt0i08cXlzbuxcr8htTTFShf8NgY2/wYDhUMnXTNfyCkj7Gtb+AnXP8TajH2wa6jATTT9rsO3cf5C5vmae2at38GvGTrJyXNNOjXIlaFenPKl1y3Nm7fI0rZp82lM1HFEYVm1nxsptbN17EDh+YcjKyeO2j+YwZVkm/+jTkkFBWLzcnKJpr7rO4S6PwPn3/f74wX3wfEM4oz9c9m/v8hVg01CbiJOXrxzIyePAQfe1Pyf38O0DOXnsP+Z27uHbe7NzWbRhN+lb9gIQFyO0qF6GAR1qk1qnAu3qlKdq2cAvYCIix5wxrMjcy7SV25m+chs/Ls9klO+MoWqZpMNF4avfNvLziq38X7/WXN2+VsBzmVO04nuY9Ag0uxzOvffI5xJKQZMesGg09Pg/iPVmlTZ/WSEwISk7N49Xvk1n5qrt7M/JZf/BPLIO5rHf92Z+qFP2ZJSIj6VkQiwlEmJpVKU0fdvW4Mza5TmjVllPrlwVERpWSaZhlWSu61jnmMLwU/pWRs/fQIzAv646gyvOrFnsGU0Rtq+CETdB5abQ543C+wFa9nMdxiunQKOuxZ/xJERMIVDViB9HHW7NeKdqZeZe7ho+j4Xrd9OuTnkql06kZEIcJRJij3gz//12XKGPJ/m+l0yIIzEuJuSnXSi8MOwDoGEVW2QlZBzcB8MHguZD/08gsYh/m4ZdIbGsKwbRXAhEpDvwHyAWeFtVnz3q+X8DXXx3SwJVVLXcyR4nKSmJbdu2UbFixYgtBqrKtm3bSEqK3DVXVZUv567nsTELSYiLYeh17bi4RVWvY3nGFQYrACFFFUbfDplLYOAXUOE4I7fiEl2z0ZKxkPNviA/dv92gFQIRiQVeBboBGcAsERmrqosPbaOqdxfY/k6g7akcq2bNmmRkZJCZmXmaqUNbUlISNWtGZvPAnqwcHh29kNHzN3BWvQq81L8N1cqW8DqWMUf66UVYPBq6Pek+8Z9Iyytg/seQ/o0rCiEqmGcEHYB0VV0JICLDgd7A4iK2HwA8fioHio+Pp149G04Xruav28ldw+aRsWM/93RrzJ+7NLT55k3oWTYJvn0KWl4JZ9/l32vqnQ8lK8GCEVFbCGoA6wrczwDOKmxDEakD1AO+K+L5wcBggNq1awc2pfFMfr4y9MeVvDAxjZQySXx+WydS61bwOpYxx9qaDl/eClVbQq9X/J9DKDYOWvSBeZ9A9h5ITA5qzFMVKpe89QdGqGpeYU+q6lBVTVXV1MqVKxdzNBMMW/ZkccN7M3l2/FK6NU9h3F3nWhEwoSlrt7tyODYO+n8KCSVP7vUtr4TcA5A2ITj5AiCYZwTrgYKDnmv6HitMf+DPQcxiQsgPaVv46xe/sicrl2f6tmJAh1oR28lvwlx+Poy6Dbalw/Wj3fQRJ6vWWVCmBiwcAa2vCnjEQAhmIZgFNBKRergC0B+49uiNRKQpUB6YFsQsJgQczM3n+YlLeevHVTRJSebTP3SkcUponiobA8CU5yBtHHR/Duqdd2r7iImBFn1hxpuwfzuUDL0z36A1DalqLnAHMBFYAnyuqotE5EkR6VVg0/7AcI2WQfJRatXWffR7/Rfe+nEV13Wsw5g7OlsRMKFtyVcw5VloMxDOOnY24ZPS6krIz4El/wtMtgCLiLmGTGgbOTeDR0cvJC42huf6taZ7y+i9NsCEiS1L4O2uUKkx3DT+9K8BUIVXzoSyteCGsYHJeJJsriHjib3ZuTw6eiGj5q2nQ113bUD1cnZtgAlxB3a4zuH4knDNx4G5EEzEdRr/+ALs2QzJKae/zwAKlVFDJsL8lrGTni//yJj567m7a2OGDe5oRcCEvvw8N0x05zq45iMoWyNw+27Zz01LsXh04PYZIFYITEDl5ytDp67gitd+ISc3n+GDOzGkayO7QMyEh++egvTJcOnzULtjYPddpSlUaeHmHgox1jRkAiZzTzb3fvErU5dlckmLFJ7r15pyJRO8jmWMfxZ+CT/9G9rdBKk3BecYrfrBt0/CzrWnNhQ1SOyMwATElGWZ9PjPVGas3MY/+rTkjUHtrAiY8LFpAYz+M9Tq6NYPCJYWV7jvC0cG7xinwAqBOS0Hc/N5ZtwSbnh3JhVKJTD2jnMY1LGOXSBmwse+ba5zuER5uPpDiAviB5gK9aBGasg1D1khMKds9dZ9XPnGLwydupJrz6rN2DvOoUlVuzbAhJG8XBhxoxvJ0//j4hnN07IfbPoNti4P/rH8ZIXAnJLR89bT8+UfWb11H28MOpNn+rYiKT7W61jGnJxvHoVVU926wjXaFc8xW/QFJKTOCqyz2JyUvdm5PDZmISPnrqd93fK81L8tNWxYqAk3u9bDjDdg+mtw1h+h7cDiO3aZalD3HDc19fkP+D+TaRBZITB+W5CxizuHzWXt9v0MuagRd17YkLhYO6k0YSI/H1ZNgVlvQ9p4N6a/5ZVw8T+KP0vLK+Cru10ndbXWxX/8o1ghMCeUn6+8+/MqnpuwlEqlE/n0Dx3pWL+i17GM8c+BHTB/GMx+x80iWqICnH2nGyJavq43mZr1hnH3ueYhKwQm1G3dm829n//KlGWZdGuewv/1a035UjYs1ISBDfNg1juuCSb3ANTsAH2HQvPe3q8fXKoi1O/ihpF2fcLz5iErBKZIPy7P5O7PfmV3Vg5P9W5hw0JN6Ms5AItGueaf9XPcfEGtr4b2t0C1M7xOd6RWV7q1DjJmQa0OnkaxQmCOkZOXzwuT0nhzykoaVSnNx7d2oGnVMl7HMqZo21fC7Hdh3seuKahSY7eGwBn9oUQ5r9MVrsmlEJvozlisEJhQsnbbfu4cPo9f1+1kQIfaPHZZc0ok2LBQE4Ly82D5JPfpP30ySCw0uwza3wp1z/W8ueWEkspA44vdGUz3f0KMd39nVgjMYWPmr+dvoxYSI/DawDO5tFU1ryMZc6y9mTDvQ5j9HuxaB8nV4IKH4MzroUx1r9OdnJZXusVqVv8E9c/3LIYVAsO+7FweH7uIEXMySK1Tnpf6t6Fm+ZNcoNuYYFKFtdPdp//FY9xqX/XOg0ue9jWxxHud8NQ0vgQSSrv1jK0QGK8sXL+Lu4bNY9W2fdx1YUPuuqiRXRsQ6Q7uh6nPu3lvzrze6zQntm0FfH4DbF4AiWVcx2/qzVC5idfJTl98CWjaExaPhUv/Fdx5jo7DCkGUUlXe/Xk1z41fSvlS8Xx6a0c6NbBrAyJexmw3UmVbOsTEQbU2ITGOvUj5eTDqj64J6PL/QKurIKGU16kCq2U/+O0zWPEdNOnuSQT76BeFtu3N5ub3Z/HUV4s5r3Flxg85z4pApMvLge+ehncuhtxstwRjyYow5nb3XKia8QZkzHRTQ7e7MfKKALjrCZLKeTr3kJ0RRJmf07dy92fz2Xkgh7/3asH1nezagIi3ZSmMGgwbf4U2A90IlaSygMBnA91iLOff73XKY21bAd8+BY27u2sBIlVcgrvIbeGXrtkuofj754J6RiAi3UUkTUTSReTBIra5WkQWi8giEfk0mHmi2cHcfP45fgmD3plBclIco2/vzA1n17UiEMny82Haq/DmebArw50F9HnNVwRwQy1b9oMp/webF3ub9Wj5+TD2TohNcDODRvr/05b94OBeWD7Rk8MH7YxARGKBV4FuQAYwS0TGquriAts0Ah4COqvqDhGpEqw80Sx9yx6GDJ/Pog27GdChNo9e1oySCXYyGNF2roXRt8PqH6FxD+j1MpQu5M+rx//ByimuieiWyRAbIv8vZr8Da36GXv8NvyGhp6LuOVA6xZ0VtOhb7IcP5hlBByBdVVeq6kFgOND7qG3+ALyqqjsAVHVLEPNEHVXlw2mr6fnyT2zclcXQ69rxzytaWRGIZKow/1N4vbOba6fXf2HAsMKLAECpSm6h9g3zYNorxZu1KDtWwzePQ4OLoO0gr9MUj5hYVwCWTYKsXcV/+CDuuwawrsD9DN9jBTUGGovIzyIyXUQK7TIXkcEiMltEZmdmZgYpbmTJ3OM6hB8bs4iO9Ssy4S/ncnGLql7HMsG0byt8NghG/wmqtoI//QxnXnfiZpUWfaHZ5fD9PyFzWfFkLYoqjL0LJMaNEor0JqGCWvaDvGxYOq7YD+31qKE4oBFwATAAeEtEyh29kaoOVdVUVU2tXLly8SYMQ98u2Uz3l6by84ptPHF5c96/qT1Vkj2ebdEE19Jx8FpHN+XCxf+AG77yf4plEej5ouukHHO7G7LplTnvuzUDLn4SytXyLocXaraHsrU9GT0UzEKwHij4L1nT91hBGcBYVc1R1VXAMlxhMKfgwME8/jZqAbd8MJvKyYl8dec53Ni5nnUIR7Ks3TDmzzB8ACRXhcFT3Fz7MSf5p126iusvyJjlhmx6YVcGTHrUXTHc7iZvMnhJxC1Ys/J72LetWA8dzEIwC2gkIvVEJAHoD4w9apvRuLMBRKQSrqloZRAzRawFGbvo+cqPfDJjLYPPq8+YOzrTOMUWko9oq3+GNzq7PoFz/wq3fgcpzU99f62uckM1v33KDd0sTqrwvyGgeXD5y9HVJFRQqyshPxeWjCnWwwatEKhqLnAHMBFYAnyuqotE5EkR6eXbbCKwTUQWA98D96lq8ZbCMJeXr7z2Qzp9X/uZ/dl5fHrrWTx8aTMS42zG0IiVkwUT/wbv93RXB988ES569PSnJxBxQzVjE9zQzfz8wOT1x/xP3QyiXZ9wU19Eq5SWbgrtBcXbPCSqWqwHPF2pqak6e/Zsr2OEhPU7D3DPZ/OZsWo7PVtV4+m+LSlX0lYPi2gbf4WRt0HmEki9BS5+KvBX28772DU39Xgezhoc2H0XZvdGeO0sqNICbvz65Ju1Is0Pz7qvexYHdOisiMxR1dTCnovy33j4GjN/Pd1fmsrC9bt44aoz+O+1ba0IRLK8XJj6Arx1kVt4ZeCXcNmLwZlyoc1AN3Rz8hNuKGcwqbpF3HOzofd/rQiAGz2EwqLRxXZI+62Hmd1ZOfxl+DyGDJ9PoyqlGTfkXK5sV9M6hCPZthXwXg/47ik3zPP2adCoa/COJ+IbuhnjhnIGs9VgwQhYNh4ufBQqNgjeccJJpUZQtbWbmrqY2JVFYWTmqu3c/dl8Nu3O4i9dG3FHl4Y2ZXQkUYW9m92yi9tXwY5V7nvaODfffr93XGdicShXyw3h/OpuN6QzNQijePZugfH3uWGTHf8U+P2Hs1ZXwjePuX//YugzsUIQBnLy8nlp8jJe/2EFNcuX5Is/duLM2uW9jmVORV6um1J5+8rf3+gPvenvWA05+3/fVmKgbC1o0sNdG1DcUy20u8ktozjpUWjYNfDj+r++102y1vtVT5dpDEkt+rpCsPBLOO+vQT+cFYIQtzJzL3/5bD6/ZeziqnY1ebxXC0on2j9bSMs54N7Uj/5kv32lKwL5ub9vG5fkLvwqX89NR1yhnrtdoR6Uq+3tylsi0OsVeO1s+OovMHBE4IZ1LhoNS8bCRY9HxgIzgVauNtQ6CxaOtEIQzVSVYTPX8dRXi0mIi+H1gWfSw9YQDl27N7gFVLYugz0bj3wusax7Y6/exl0wdOiNvkJ9KF01tDtIy9d1QzrH3+eGeLYdePr73LfNnQ1UawNn33X6+4tULa90v/ctS6BKs6AeygpBiFFV5q7dwSvfpfNDWiadG1bkX1e1oWpZmyIipP32uZsaoXV/1+lZof7vb/glyof3BVLtb3VNRBMfggYXQpnT/EAy/n43sdoNY0NnttNQ1KIPTHjANQ9d+EhQD2X/CiHiYG4+4xdu5N2fVvFrxi7KJMXxSM9m3Ny5HjExYfwmEi2WTXATvV3xptdJAi8mxg3tfP1s13k8YNipF7alX7vRMBc8DCktApsz0pSu4qbbWPgldPlbUD9MWCHw2PZ9Bxk2cy0fTlvN5t3Z1K9ciqf6tKTfmTVsuuhwsW8brJvhpnmIVBUbuCGek/7mhny2vurk93FghyskKa3g3HsCnzESteznrvLeMA9qnBm0w9g7jUfSNu3hvZ9XMWreerJz8zm3USWe7dea8xtVtjOAcLN8Emi+G90TyTr+CRaPce3W9c8veo2Dokx4GPZvg4FfeNsJHk6aXQ5f3ePOCqwQRIb8fOWHZVt496fV/JS+laT4GPq1q8lNZ9elkU0QF76WjXedvtXaeJ0kuGJi3VDPN85xnb3XfOT/a5d/A79+CufdB9XOCF7GSFOivBu6u2gUdHsqaAMLrBAUg33ZuYyYk8H7v6xm1dZ9VC2TxP3dmzCgfW3Kl7JpIcJabjakf+dGA4Xy6J9AqdwYLngQvv27e3PyZ1nFrF3uCuXKzVwhMCenZT/3YWPddKhzdlAOYYUgiNZt38+H01YzfNY69mTl0qZWOV4e0JYeLasSb1cER4bVP8HBPdDkUq+TFJ+z73LXAHz9V6h7HpSqePztJz0KezdB/48hLrF4MkaSJj0groRrHvKyEIjISOAdYLyqFuPctOFHVZm1egfv/rSKSYs3ISJc2qoaN3Wua1cDR6JlE9wfaf3zvU5SfGLjoPdr8OZ5bijole8Uve2K72HuB9B5CNRoV3wZI0liaWjS3V2E1/25oAy59XePrwE3AS+LyBfAe6qaFvA0YSw7N4+vf9vIuz+vYuH63ZQtEc9t5zfguo51qF6uhNfxTDCoQtoEaNAF4qPs3zilOZx/P3z/tGsWa9rz2G2y97gmoYqN3HBRc+pa9nNNcaumQMOLAr57vwqBqk4GJotIWdzawpNFZB3wFvCxquYEPFmY2Ln/IB9OW8NH09eQuSebhlVK83TfllzRtiYlEmz+lIi2eRHsWlssUwCEpHPudk1EX90NtTtByQpHPj/5CTelxs0TId4uiDwtDbtB9bZu+pIg8PscQ0QqAoOA64B5wCfAOcAN+JabjDY5efkMeGsGSzbu5oImlbm5cz3ObVTJpoSOFsvGu++NL/E2h1di490oorcuhIkPQ98Cax2v+hFmvQ0db4faZ3mXMVLEJ8HgH4K2e3/7CEYBTYCPgMtV9dBkKp+JSNQuF/bWjytZsnG3zQMUrdLGu3bv5KpeJ/FOtTPcmcHU56HFFdD4Yji4D8be4abYuPBRrxMaP/g7dOVlVW2uqv8sUAQAKGrps0i3eus+/jN5Od1bVLUiEI32bIb1c6BxhF9E5o/z7nNDQ/83xA0V/e4fbvbV3v+FhJJepzN+8LcQNBeRcofuiEh5Ebk9OJFCn6ryt9ELSIiN4YleNl9KVFo+0X1v0t3bHKEgLhH6vOqGiA4fCNNfh/Z/gLrneJ3M+MnfQvAHVd156I6q7gD+EJREYWDk3PX8nL6N+3s0tVlBo1XaeLdoTEpLr5OEhhrt4Ow7YfWP7vfS9QmvE5mT4G8hiJUCPaAiEguc8JJYEekuImkiki4iDxby/I0ikiki831ft/of3Rvb9mbzj68X065OeQZ2qO11HOOFnANufHzj7uE9vXSgXfCwW9Xsqvfc2HcTNvwdNTQB1zF8aI7d23yPFclXLF4FugEZwCwRGauqi4/a9DNVveMkMnvq6a+XsDc7l39e0comh4tWq6ZC7gFrFjpafBJc/pLXKcwp8LcQPIB78z+0wvQ3wNsneE0HIF1VVwKIyHCgN3B0IQgbPy7PZOS89dx5YUMa2yRx0SttHCSUhrrnep3EmIDw94KyfOB135e/agDrCtzPAAobUNxPRM4DlgF3q+q6ozcQkcHAYIDatb1pjjlwMI+/jVpIvUql+HOXhp5kMCFAFZZNdCt12bw5JkL41UcgIo1EZISILBaRlYe+AnD8/wF1VbU17izjg8I2UtWhqpqqqqmVK1cOwGFP3svfLWft9v0807cVSfF2xXDU2jjfrUkcTZPMmYjnb2fxe7izgVygC/Ah8PEJXrMeqFXgfk3fY4ep6jZVzfbdfRsIyVmplmzczdCpK7mqXU06NTjBTIsmsqWNB4mBRhd7ncSYgPG3EJRQ1W8BUdU1qvoEUMgsU0eYBTQSkXoikgD0B8YW3EBECl6J1QtY4meeYpOXrzw4cgHlSsTz8KXNvI5jvJY2Hmp2OPHUy8aEEX87i7NFJAZYLiJ34D7ZH3d8mKrm+radCMQC76rqIhF5EpitqmOBu0SkF+5MYztw4yn+HEHz0bTV/LpuJ//p38YWkYl2uzJg02/Q9e9eJzEmoPwtBEOAksBdwFO45qEbTvQiVR0HjDvqsccK3H4IeMjfsMVtw84DPD8xjfMaV6bXGdW9jmO8tsw3YjrS1yY2UeeEhcB3PcA1qvpXYC9uXYKIp6o8NmYheao83aelzShq3NoD5etBpcZeJzEmoE7YR6CqebjppqPKhIWbmLxkC/d0a0ytCjZxVtTL3usWBWlyqV1NbCKOv01D80RkLPAFsO/Qg6o6MiipPLbrQA6Pj11E82pluLlzPa/jmFCw8nvIO2hXE5uI5G8hSAK2ARcWeEyBiCwE/zdhKVv3ZvP2DanE2SLzBlyzUGJZtxKXMRHG3yuLo6JfAGDW6u18MmMtt5xTj9Y1y3kdx4SC/DzXUdyom1uVy5gI4+8KZe/hzgCOoKo3BzyRh7Jz83ho5AJqlCvBPd2sQ9D4rJ8D+7faaCETsfxtGvqqwO0koC+wIfBxvPXmlJWkb9nLeze2p1Si38s5m0iXNg4kFhpe5HUSY4LC36ahLwveF5FhwE9BSeSRFZl7+e936VzWuhpdmlbxOo4JJWkToM7ZUKK810mMCYpT7QltBETMu2V+vvLQyAUkxcfw2OXNvY5jQsn2VZC5xJqFTETzt49gD0f2EWzCrVEQEb6Ys46Zq7bz7BWtqJJsS0+aAuxqYhMF/G0aithVWDL3ZPP010voUK8CV6fWOvELTHRJGw+VmkCF+l4nMSZo/F2PoK+IlC1wv5yI9AlaqmL05FeLycrJ55m+tvSkOUrWLljzs11EZiKev30Ej6vqrkN3VHUn8HhQEhWj79O28L9fN/DnLg1pWMUW2zZHSZ8M+bm2CI2JeP4WgsK2C+vxlfsP5vLIqIU0rFKaP15gp/2mEGkToGRFqNne6yTGBJW/hWC2iLwoIg18Xy8Cc4IZLNj+/c0y1u88wD+vaEVinC09aY6SlwvLJ7mVyGLs/4eJbP4WgjuBg8BnwHAgC/hzsEIF28L1u3jnp1Vce1Zt2tet4HUcE4rWTYesnTZayEQFf0cN7QMeDHKWYpGbl8+DI3+jYulEHuje1Os4JlSljYfYBGhw4Ym3NSbM+Ttq6BsRKVfgfnkRmRi0VEH0/i+rWbh+N09c3oKyJWwCMVOEtPFQ9xxIjNiR08Yc5m/TUCXfSCEAVHUHYXhl8brt+/nXpGVc1LQKl7aq6nUcE6q2LoftK2y0kIka/haCfBGpfeiOiNSlkNlIQ5mq8sjohYjAk7b0pDmetPHue+NLvM1hTDHxdwjo34CfRGQKIMC5wOCgpQqC//22kSnLMnnssubUKFfC6zjGHz/9230/5+7iPW7aeEhpCeVqn3hbYyKAX2cEqjoBSAXSgGHAvcCBE71ORLqLSJqIpItIkZ3NItJPRFREUv3MfdIqlEzg0lZVueHsusE6hAmkxWNh8hPua+5HxXfc/dvdiCEbLWSiiL+Tzt0KDAFqAvOBjsA0jly68ujXxAKvAt2ADGCWiIxV1cVHbZfs2/eMU8jvt3MaVeKcRpWCeQgTKLsyYOydUK0NlCgHX98DVZpBzaB9Tvjd8m9A86GxFQITPfztIxgCtAfWqGoXoC2w8wSv6QCkq+pKVT2Iu/6gdyHbPQU8h7s2wUS7/Dz48g9uaocr34Ur34PkavDZINizKfjHTxsHpVOgetvgH8uYEOFvIchS1SwAEUlU1aVAkxO8pgawrsD9DN9jh4nImUAtVf36eDsSkcEiMltEZmdmZvoZ2YSlqS/A2l/g0hegYgMoWQEGDHMTwH12HeRmB+/YuQch/VvXSRxzqkt1GBN+/P3fnuG7jmA08I2IjAHWnM6BRSQGeBHX33BcqjpUVVNVNbVy5cqnc1gTytZOhynPQqur4Yz+vz+e0gL6vA4ZM2HcX0GDNGBtzc9wcI81C5mo4++VxX19N58Qke+BssCEE7xsPVBwgv+avscOSQZaAj/4hnJWBcaKSC9Vne1PLhNBDuyAL291I3V6/guOHt7bog9suhd+/JfrO2h/S+AzpI2HuCSof0Hg921MCDvpGURVdYqfm84CGolIPVwB6A9cW2A/u4DDvbci8gPwVysCUUgV/jcE9myEmydBUpnCt+vyN9i0AMbf7zqP65wd2AzLxrsikFAycPs1JgwErSFUVXOBO4CJwBLgc1VdJCJPikivYB3XhKG5H8LiMXDhI1CzXdHbxcTCFW9B+brw+fVudFGgbFkMO9dCY1uExkSfoPaIqeo4VW2sqg1U9WnfY4+p6thCtr3AzgaiUGYajH/AfRI/e8iJty9RDvp/CjlZMHwg5Jzwchb/HL6a2AqBiT42NMJ4JycLRtzimmL6vun/SJ3KTeCKobBxPnx1d2A6j5dNcENGy1Q7/X0ZE2asEBjvTH4cNi9wI4KST3ISwKaXuj6DX4fB9NdPL8feLZAx20YLmahlhcB4I20CzHgDzvrTqU/udu5foellMOkRWPnDqWdZNhFQm1bCRC0rBKb47d4IY26Hqq2g299PfT8xMdD3DajUGL64CXasPrX9LJsAZWq6PMZEISsEpnjl58Oo21wnb793IS7x9PaXmAz9PwHNg+GD4OC+k3t9Thas+A6adD/22gVjooQVAlO8fvkPrJoCPZ6Dyo0Ds8+KDdy8RFsWwZg7Tq7zeNVUyNlv/QMmqlkhMMUnYzZ89w9o0RfaXhfYfTfsChc9DotGws8v+f+6tHEQX8otS2lMlLJCYIpH1m4YcTMkV4fLXgpOM0znIdCyH0z+OyyffOLtVV1HccMLIT4p8HmMCRNWCEzwqbo1BXZlQL+33UVhwSACvf4LVVu6orNtxfG33/gr7NlgzUIm6lkhMMH363BY8AVc8BDUPiu4x0ooCdd84qajGDYAsvcUvW3aeECg0cXBzWRMiLNCYIJr2wr4+l6ocw6ce0/xHLN8Hbj6A9iWDqP+6EYqFWbZeKjVAUrb1OYmulkhMMGTe9A10cQluCkhYmKL79j1zoNLnoGlX8HU5499ftd61zRkcwsZc/LTUBvjt++edPMBXfMJlK1xws0D7qzb3Jv9D8+4foOmPX9/bplvOQ27mtgYOyMwQZL+LfzyCqTeAs0u8yaDCFz2b6h+JowcDFuW/v7csgluOuvKTb3JZkwIsUJgAm/vFtc2X7kZXPK0t1nik+CajyG+JAy/Fg7sdFcfr5ziRgvZ1cTGWCEwAZafD6P/BNm73dW+8SW8TuSapa75yC088+Wt7mwlL9tNK2GMsT4CE2AzXof0ydDzRUhp7nWa39XuCJf+n1u/YP0cSCwLdTp7ncqYkGBnBCZwNsyHbx53U0On3ux1mmOl3gztboQD26HhRRAb73UiY0KCnRGYwMje64aKlq4CvV4J3bb3Hs9DXBK0vtrrJMaEDCsEJjDGPwA7VsEN/4OSFbxOU7S4BDfzqTHmMGsaMqdv8ViY/7FbMcxm8TQm7AS1EIhIdxFJE5F0EXmwkOf/KCILRGS+iPwkIiHUu2j8ogo/vuBWCTv/Aa/TGGNOQdAKgYjEAq8CPYDmwIBC3ug/VdVWqtoG+D/gxWDlMUGybqa7evesP0KstTQaE46CeUbQAUhX1ZWqehAYDvQuuIGq7i5wtxRwEktLmZAw8003FLP1NV4nMcacomB+hKsBrCtwPwM4Zg5iEfkzcA+QAFxY2I5EZDAwGKB27doBD2pO0e6NsHgMdLgNEkt7ncYYc4o87yxW1VdVtQHwAPBIEdsMVdVUVU2tXNmmDA4Zc96D/DzocKvXSYwxpyGYhWA9UKvA/Zq+x4oyHOgTxDwmkHKzYfa7blGXCvW9TmOMOQ3BLASzgEYiUk9EEoD+wNiCG4hIowJ3ewLLg5jHBNKi0bAv0031bIwJa0HrI1DVXBG5A5gIxALvquoiEXkSmK2qY4E7RKQrkAPsAG4IVh4TYDPfhIqNoH4Xr5MYY05TUMf7qeo4YNxRjz1W4PaQYB7fBEnGHDdxW4/nIcbzbiZjzGmyv2Jz8ma+CQnJ0GaA10mMMQFghcCcnD2bYeFIaHMtJCZ7ncYYEwBWCMzJmfM+5OdAh8FeJzHGBIgVAuO/3INuyGjDrlCpoddpjDEBEl2FIC/X6wThbclY2LvJXUlsjIkY0VMIfvschp4PWbu8ThK+Zg51F4817Op1EmNMAEVPIShTAzKXwsjBboF1c3I2zIN1M6D9H2zIqDERJnr+out2hu7PwrIJ8MMzXqcJPzOGQnwpaDvQ6yTGmACLnkIA0P5WaDsIpj7vZs00/tm3FRZ+6a4bSCrrdRpjTIBFVyEQgZ4vQs32MOpPsHmx14nCw5z3IS/bhowaE6GiqxAAxCXC1R+5+fOHD4D9271OFNrycmDWO1D/AqjcxOs0xpggiL5CAFCmGlzzMezeAF/eYsNKj2fpV7Bngw0ZNSaCRWchAKjVAS59AVZ8B9/+3es0oWvGUChXBxpf4nUSY0yQRG8hAGh3g+tA/uVlWDDC6zShZ9MCWPsLdPgDxMR6ncYYEyTRXQgALvkn1D4bxtwBG3/1Ok1omfEmxJd0I62MMRHLCkFcAlz9AZSsAMMHuqGSxnWiL/gCWl8NJcp7ncYYE0RWCABKV4H+n7ilF7+40Y2UiXZzP4DcLBsyakwUsEJwSPW2cPl/YPWPMOkRr9N4Ky/XDRmtey6ktPA6jTEmyKwQFHRGf+j4Z5jxBsz72Os03lk2Hnats4XpjYkSVgiO1u1JqHc+fHU3ZMz2Oo03ZrwJZWtB4x5eJzHGFAMrBEeLjYOr3ofkqvDZILc0YzTZvMg1j7W/xf0ujDERL6iFQES6i0iaiKSLyIOFPH+PiCwWkd9E5FsRqRPMPH4rWQH6f+rWLvj8OrcyV7SYORTikuDMG7xOYowpJkErBCISC7wK9ACaAwNEpPlRm80DUlW1NTAC+L9g5TlpVVtB71fdHPzj7/M6TfE4sMMt4NPqKlcMjTFRIZhnBB2AdFVdqaoHgeFA74IbqOr3qrrfd3c6UDOIeU5eyyvgnLvd7Juz3/U6TfDN+xhy9lsnsTFRJpiFoAawrsD9DN9jRbkFGF/YEyIyWERmi8jszMzMAEb0w4WPQsNuMO5+WDOteI9dnPLzYOZb7irrqq28TmOMKUYh0VksIoOAVOD5wp5X1aGqmqqqqZUrVy7ecDGx0O9tKFcbPr8edq0v3uMXl2UTYecaOMsuIDMm2gSzEKwHahW4X9P32BFEpCvwN6CXqmYHMc+pK1HOdR7n7HcjiXKyvE4UeDPfhOTq0PQyr5MYY4pZMAvBLKCRiNQTkQSgPzC24AYi0hZ4E1cEtgQxy+mr0hSuGAob5rprDFS9ThQ4mWmw8gffkNF4r9MYY4pZ0AqBquYCdwATgSXA56q6SESeFJFevs2eB0oDX4jIfBEZW8TuQkPTnnDBQ/Drp+6iq0gxcyjEJkK7G71OYozxQFCvGFLVccC4ox57rMDtrsE8flCcdz9s/A0mPgxVmkH9871OdHqydsH8YdCyH5Sq5HUaY4wHQqKzOKzExEDfN6BiQzdT6Y41Xic6PfM+gZx91klsTBSzQnAqksrAgGFuyOXwgXBw/4lfE4ry82HWW1Czg5t91RgTlWwymVNVsQFc+Q58chX8NxUSywR2/yJQu6NbSjNYU0GnT4btK6HL34Kzf2NMWLBCcDoadYMr3oIlQejjzs2G+Z+6K5prd3IFoVkvt6JaoMx8E0pXhea9T7ytMSZiWSE4Xa2vcl/BsH+7rxi8A1/eAqUqw5nXQ7uboFytE7/+eLamuzOCCx62IaPGRDnrIwhlJSvA2XfAHXNg0EjXlv/Tv+E/rWHYAPdGnp9/avueORRi4m3IqDHGzgjCQkwMNLzIfe1c5ybBm/sBpI2D8vUg9WZoO8j/GUOz97gzjRZ9ITklqNGNMaHPzgjCTblacNGjcPdi6PcOJFeDbx6FfzWFUX+CjDkn3sf8YXBwD5z1x+DnNcaEPDsjCFdxCdDqSve1eZFbbP63z9xVz9XauM7llv0goeSRr8vPd81CNdpBzXaeRDfGhBY7I4gEKS3gshfh3qXQ819uxNHYO+DFpjDhYdcxfMjK72Hbcuhgaw4YYxzRMJs8LTU1VWfPjtJF5f2lCmunway3YfFYyM+B+l3cWcKc92HjfLh7EcQlep3UGFNMRGSOqqYW9pw1DUUiEahztvvasxnmfQiz34fPBrrnz7vfioAx5jArBJEuOQXOuw863w3LJ8KK76Djn7xOZYwJIVYIokVsnJtGu2lPr5MYY0KMdRYbY0yUs0JgjDFRzgqBMcZEOSsExhgT5awQGGNMlLNCYIwxUc4KgTHGRDkrBMYYE+XCbq4hEckE1pziyysBWwMYJ9jCKW84ZYXwyhtOWSG88oZTVji9vHVUtXJhT4RdITgdIjK7qEmXQlE45Q2nrBBeecMpK4RX3nDKCsHLa01DxhgT5awQGGNMlIu2QjDU6wAnKZzyhlNWCK+84ZQVwitvOGWFIOWNqj4CY4wxx4q2MwJjjDFHsUJgjDFRLmoKgYh0F5E0EUkXkQe9zlMUEaklIt+LyGIRWSQiQ7zO5A8RiRWReSLylddZjkdEyonICBFZKiJLRKST15mOR0Tu9v0/WCgiw0QkyetMBYnIuyKyRUQWFnisgoh8IyLLfd/Le5nxkCKyPu/7v/CbiIwSkXIeRjyssKwFnrtXRFREKgXqeFFRCEQkFngV6AE0BwaISHNvUxUpF7hXVZsDHYE/h3DWgoYAS7wO4Yf/ABNUtSlwBiGcWURqAHcBqaraEogF+nub6hjvA92PeuxB4FtVbQR867sfCt7n2KzfAC1VtTWwDHiouEMV4X2OzYqI1AIuBtYG8mBRUQiADkC6qq5U1YPAcKC3x5kKpaobVXWu7/Ye3BtVDW9THZ+I1AR6Am97neV4RKQscB7wDoCqHlTVnZ6GOrE4oISIxAElgQ0e5zmCqk4Fth/1cG/gA9/tD4A+xZmpKIVlVdVJqprruzsdqFnswQpRxO8V4N/A/UBAR/lESyGoAawrcD+DEH9zBRCRukBbYIbHUU7kJdx/znyPc5xIPSATeM/XjPW2iJTyOlRRVHU98ALu099GYJeqTvI2lV9SVHWj7/YmIMXLMCfhZmC81yGKIiK9gfWq+mug9x0thSDsiEhp4EvgL6q62+s8RRGRy4AtqjrH6yx+iAPOBF5X1bbAPkKn2eIYvrb13rgCVh0oJSKDvE11ctSNTw/5Meoi8jdcs+wnXmcpjIiUBB4GHgvG/qOlEKwHahW4X9P3WEgSkXhcEfhEVUd6necEOgO9RGQ1rsntQhH52NtIRcoAMlT10BnWCFxhCFVdgVWqmqmqOcBI4GyPM/ljs4hUA/B93+JxnuMSkRuBy4CBGroXVjXAfSD41fe3VhOYKyJVA7HzaCkEs4BGIlJPRBJwHW5jPc5UKBERXBv2ElV90es8J6KqD6lqTVWti/u9fqeqIfmpVVU3AetEpInvoYuAxR5GOpG1QEcRKen7f3ERIdy5XcBY4Abf7RuAMR5mOS4R6Y5r1uylqvu9zlMUVV2gqlVUta7vby0DONP3f/q0RUUh8HUG3QFMxP0hfa6qi7xNVaTOwHW4T9bzfV+Xeh0qgtwJfCIivwFtgGe8jVM035nLCGAusAD39xpSUyKIyDBgGtBERDJE5BbgWaCbiCzHndU862XGQ4rI+l8gGfjG97f2hqchfYrIGrzjhe6ZkDHGmOIQFWcExhhjimaFwBhjopwVAmOMiXJWCIwxJspZITDGmChnhcCYYiQiF4T6DK0m+lghMMaYKGeFwJhCiMggEZnpu8joTd96C3tF5N++9QG+FZHKvm3biMj0AnPal/c93lBEJovIryIyV0Qa+HZfusCaCJ/4rho2xjNWCIw5iog0A64BOqtqGyAPGAiUAmaragtgCvC47yUfAg/45rRfUODxT4BXVfUM3BxBh2bkbAv8Bbc2Rn3c1eTGeCbO6wDGhKCLgHbALN+H9RK4idPygc9823wMjPStcVBOVaf4Hv8A+EJEkoEaqjoKQFWzAHz7m6mqGb7784G6wE9B/6mMKYIVAmOOJcAHqnrEalUi8uhR253q/CzZBW7nYX+HxmPWNGTMsb4FrhSRKnB4Dd46uL+XK33bXAv8pKq7gB0icq7v8euAKb7V5TJEpI9vH4m+OeWNCTn2ScSYo6jqYhF5BJgkIjFADvBn3EI2HXzPbcH1I4CbavkN3xv9SuAm3+PXAW+KyJO+fVxVjD+GMX6z2UeN8ZOI7FXV0l7nMCbQrGnIGGOinJ0RGGNMlLMzAmOMiXJWCIwxJspZITDGmChnhcAYY6KcFQJjjIly/w+hw2DXC1pbDwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_loss_accuracy(history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "flDHUqTqREin"
   },
   "source": [
    "#### Model 5: Conv 3D Model with 30 frames per video (8, 16, 32, 64 filters conv 3D layers + 256 dense layer + 128 dense layer + image size 120 by 120)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "QTtsxjdwNr_2"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# epochs = 25\n",
      "# batch size = 10\n",
      "# num_frames per video = 30\n",
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv3d_4 (Conv3D)           (None, 30, 120, 120, 8)   656       \n",
      "                                                                 \n",
      " activation_7 (Activation)   (None, 30, 120, 120, 8)   0         \n",
      "                                                                 \n",
      " batch_normalization_6 (Batc  (None, 30, 120, 120, 8)  32        \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " max_pooling3d_4 (MaxPooling  (None, 15, 60, 60, 8)    0         \n",
      " 3D)                                                             \n",
      "                                                                 \n",
      " conv3d_5 (Conv3D)           (None, 15, 60, 60, 16)    3472      \n",
      "                                                                 \n",
      " activation_8 (Activation)   (None, 15, 60, 60, 16)    0         \n",
      "                                                                 \n",
      " batch_normalization_7 (Batc  (None, 15, 60, 60, 16)   64        \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " max_pooling3d_5 (MaxPooling  (None, 7, 30, 30, 16)    0         \n",
      " 3D)                                                             \n",
      "                                                                 \n",
      " conv3d_6 (Conv3D)           (None, 7, 30, 30, 32)     13856     \n",
      "                                                                 \n",
      " activation_9 (Activation)   (None, 7, 30, 30, 32)     0         \n",
      "                                                                 \n",
      " batch_normalization_8 (Batc  (None, 7, 30, 30, 32)    128       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " max_pooling3d_6 (MaxPooling  (None, 3, 15, 15, 32)    0         \n",
      " 3D)                                                             \n",
      "                                                                 \n",
      " conv3d_7 (Conv3D)           (None, 3, 15, 15, 64)     55360     \n",
      "                                                                 \n",
      " activation_10 (Activation)  (None, 3, 15, 15, 64)     0         \n",
      "                                                                 \n",
      " batch_normalization_9 (Batc  (None, 3, 15, 15, 64)    256       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " max_pooling3d_7 (MaxPooling  (None, 1, 7, 7, 64)      0         \n",
      " 3D)                                                             \n",
      "                                                                 \n",
      " flatten_1 (Flatten)         (None, 3136)              0         \n",
      "                                                                 \n",
      " dense_3 (Dense)             (None, 256)               803072    \n",
      "                                                                 \n",
      " activation_11 (Activation)  (None, 256)               0         \n",
      "                                                                 \n",
      " batch_normalization_10 (Bat  (None, 256)              1024      \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " dropout_2 (Dropout)         (None, 256)               0         \n",
      "                                                                 \n",
      " dense_4 (Dense)             (None, 128)               32896     \n",
      "                                                                 \n",
      " activation_12 (Activation)  (None, 128)               0         \n",
      "                                                                 \n",
      " batch_normalization_11 (Bat  (None, 128)              512       \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " dropout_3 (Dropout)         (None, 128)               0         \n",
      "                                                                 \n",
      " dense_5 (Dense)             (None, 5)                 645       \n",
      "                                                                 \n",
      " activation_13 (Activation)  (None, 5)                 0         \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 911,973\n",
      "Trainable params: 910,965\n",
      "Non-trainable params: 1,008\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "num_epochs = 25 # choose the number of epochs\n",
    "print ('# epochs =', num_epochs)\n",
    "batch_size = 10  #experiment with the batch size\n",
    "print ('# batch size =', batch_size)\n",
    "num_frames = 30\n",
    "print ('# num_frames per video =', num_frames)\n",
    "height = 120\n",
    "width = 120\n",
    "\n",
    "conv_filters = (8, 16, 32, 64) \n",
    "model = conv3D(conv_filters, num_images=num_frames)\n",
    "print (model.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "LHDb1veZkEh1"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:`period` argument is deprecated. Please use `save_freq` to specify the frequency in number of batches seen.\n"
     ]
    }
   ],
   "source": [
    "callbacks_list = model_callbacks(\"model_conv3D_5\")\n",
    "\n",
    "steps_per_epoch, validation_steps = calculate_steps(num_train_sequences, num_val_sequences, batch_size)\n",
    "\n",
    "train_generator = generate(train_path, train_doc, batch_size, num_images=num_frames)\n",
    "\n",
    "val_generator   = generate(val_path, val_doc, batch_size, num_images=num_frames)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "7ePXVLz1Nr_6"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/25\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-08-28 16:52:32.304435: I tensorflow/stream_executor/cuda/cuda_dnn.cc:377] Loaded cuDNN version 8302\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "67/67 [==============================] - ETA: 0s - loss: 1.4916 - categorical_accuracy: 0.4495\n",
      "Epoch 00001: val_loss improved from inf to 4.91692, saving model to model_conv3D_6_2023-08-2816_51_10.362060/model-00001-1.49156-0.44947-4.91692-0.16000.h5\n",
      "67/67 [==============================] - 171s 3s/step - loss: 1.4916 - categorical_accuracy: 0.4495 - val_loss: 4.9169 - val_categorical_accuracy: 0.1600 - lr: 0.0010\n",
      "Epoch 2/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 1.1291 - categorical_accuracy: 0.5641\n",
      "Epoch 00002: val_loss did not improve from 4.91692\n",
      "67/67 [==============================] - 157s 2s/step - loss: 1.1291 - categorical_accuracy: 0.5641 - val_loss: 7.9069 - val_categorical_accuracy: 0.1500 - lr: 0.0010\n",
      "Epoch 3/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.9938 - categorical_accuracy: 0.6335\n",
      "Epoch 00003: val_loss did not improve from 4.91692\n",
      "67/67 [==============================] - 156s 2s/step - loss: 0.9938 - categorical_accuracy: 0.6335 - val_loss: 11.4103 - val_categorical_accuracy: 0.1500 - lr: 0.0010\n",
      "Epoch 4/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.7800 - categorical_accuracy: 0.6878\n",
      "Epoch 00004: val_loss did not improve from 4.91692\n",
      "\n",
      "Epoch 00004: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.\n",
      "67/67 [==============================] - 156s 2s/step - loss: 0.7800 - categorical_accuracy: 0.6878 - val_loss: 5.4066 - val_categorical_accuracy: 0.2700 - lr: 0.0010\n",
      "Epoch 5/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.6035 - categorical_accuracy: 0.7828\n",
      "Epoch 00005: val_loss improved from 4.91692 to 4.00014, saving model to model_conv3D_6_2023-08-2816_51_10.362060/model-00005-0.60347-0.78281-4.00014-0.28000.h5\n",
      "67/67 [==============================] - 155s 2s/step - loss: 0.6035 - categorical_accuracy: 0.7828 - val_loss: 4.0001 - val_categorical_accuracy: 0.2800 - lr: 1.0000e-04\n",
      "Epoch 6/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.5675 - categorical_accuracy: 0.7813\n",
      "Epoch 00006: val_loss improved from 4.00014 to 3.19170, saving model to model_conv3D_6_2023-08-2816_51_10.362060/model-00006-0.56753-0.78130-3.19170-0.23000.h5\n",
      "67/67 [==============================] - 156s 2s/step - loss: 0.5675 - categorical_accuracy: 0.7813 - val_loss: 3.1917 - val_categorical_accuracy: 0.2300 - lr: 1.0000e-04\n",
      "Epoch 7/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.5191 - categorical_accuracy: 0.8100\n",
      "Epoch 00007: val_loss improved from 3.19170 to 1.67145, saving model to model_conv3D_6_2023-08-2816_51_10.362060/model-00007-0.51906-0.80995-1.67145-0.45000.h5\n",
      "67/67 [==============================] - 157s 2s/step - loss: 0.5191 - categorical_accuracy: 0.8100 - val_loss: 1.6714 - val_categorical_accuracy: 0.4500 - lr: 1.0000e-04\n",
      "Epoch 8/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.4413 - categorical_accuracy: 0.8356\n",
      "Epoch 00008: val_loss improved from 1.67145 to 1.37638, saving model to model_conv3D_6_2023-08-2816_51_10.362060/model-00008-0.44129-0.83560-1.37638-0.49000.h5\n",
      "67/67 [==============================] - 156s 2s/step - loss: 0.4413 - categorical_accuracy: 0.8356 - val_loss: 1.3764 - val_categorical_accuracy: 0.4900 - lr: 1.0000e-04\n",
      "Epoch 9/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.5734 - categorical_accuracy: 0.7783\n",
      "Epoch 00009: val_loss improved from 1.37638 to 1.05243, saving model to model_conv3D_6_2023-08-2816_51_10.362060/model-00009-0.57343-0.77828-1.05243-0.63000.h5\n",
      "67/67 [==============================] - 155s 2s/step - loss: 0.5734 - categorical_accuracy: 0.7783 - val_loss: 1.0524 - val_categorical_accuracy: 0.6300 - lr: 1.0000e-04\n",
      "Epoch 10/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.4585 - categorical_accuracy: 0.8311\n",
      "Epoch 00010: val_loss improved from 1.05243 to 0.97562, saving model to model_conv3D_6_2023-08-2816_51_10.362060/model-00010-0.45849-0.83107-0.97562-0.64000.h5\n",
      "67/67 [==============================] - 158s 2s/step - loss: 0.4585 - categorical_accuracy: 0.8311 - val_loss: 0.9756 - val_categorical_accuracy: 0.6400 - lr: 1.0000e-04\n",
      "Epoch 11/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.4315 - categorical_accuracy: 0.8311\n",
      "Epoch 00011: val_loss improved from 0.97562 to 0.89625, saving model to model_conv3D_6_2023-08-2816_51_10.362060/model-00011-0.43145-0.83107-0.89625-0.66000.h5\n",
      "67/67 [==============================] - 160s 2s/step - loss: 0.4315 - categorical_accuracy: 0.8311 - val_loss: 0.8963 - val_categorical_accuracy: 0.6600 - lr: 1.0000e-04\n",
      "Epoch 12/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3802 - categorical_accuracy: 0.8748\n",
      "Epoch 00012: val_loss improved from 0.89625 to 0.87035, saving model to model_conv3D_6_2023-08-2816_51_10.362060/model-00012-0.38023-0.87481-0.87035-0.70000.h5\n",
      "67/67 [==============================] - 155s 2s/step - loss: 0.3802 - categorical_accuracy: 0.8748 - val_loss: 0.8704 - val_categorical_accuracy: 0.7000 - lr: 1.0000e-04\n",
      "Epoch 13/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3874 - categorical_accuracy: 0.8537\n",
      "Epoch 00013: val_loss improved from 0.87035 to 0.76808, saving model to model_conv3D_6_2023-08-2816_51_10.362060/model-00013-0.38741-0.85370-0.76808-0.71000.h5\n",
      "67/67 [==============================] - 154s 2s/step - loss: 0.3874 - categorical_accuracy: 0.8537 - val_loss: 0.7681 - val_categorical_accuracy: 0.7100 - lr: 1.0000e-04\n",
      "Epoch 14/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3226 - categorical_accuracy: 0.8763\n",
      "Epoch 00014: val_loss did not improve from 0.76808\n",
      "67/67 [==============================] - 154s 2s/step - loss: 0.3226 - categorical_accuracy: 0.8763 - val_loss: 0.8420 - val_categorical_accuracy: 0.7000 - lr: 1.0000e-04\n",
      "Epoch 15/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3004 - categorical_accuracy: 0.9035\n",
      "Epoch 00015: val_loss did not improve from 0.76808\n",
      "67/67 [==============================] - 156s 2s/step - loss: 0.3004 - categorical_accuracy: 0.9035 - val_loss: 1.0620 - val_categorical_accuracy: 0.6400 - lr: 1.0000e-04\n",
      "Epoch 16/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3041 - categorical_accuracy: 0.8793\n",
      "Epoch 00016: val_loss improved from 0.76808 to 0.64960, saving model to model_conv3D_6_2023-08-2816_51_10.362060/model-00016-0.30405-0.87934-0.64960-0.78000.h5\n",
      "67/67 [==============================] - 155s 2s/step - loss: 0.3041 - categorical_accuracy: 0.8793 - val_loss: 0.6496 - val_categorical_accuracy: 0.7800 - lr: 1.0000e-04\n",
      "Epoch 17/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2589 - categorical_accuracy: 0.9110\n",
      "Epoch 00017: val_loss did not improve from 0.64960\n",
      "67/67 [==============================] - 153s 2s/step - loss: 0.2589 - categorical_accuracy: 0.9110 - val_loss: 0.8120 - val_categorical_accuracy: 0.7100 - lr: 1.0000e-04\n",
      "Epoch 18/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2273 - categorical_accuracy: 0.9125\n",
      "Epoch 00018: val_loss did not improve from 0.64960\n",
      "67/67 [==============================] - 153s 2s/step - loss: 0.2273 - categorical_accuracy: 0.9125 - val_loss: 0.6962 - val_categorical_accuracy: 0.7600 - lr: 1.0000e-04\n",
      "Epoch 19/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2672 - categorical_accuracy: 0.9005\n",
      "Epoch 00019: val_loss did not improve from 0.64960\n",
      "\n",
      "Epoch 00019: ReduceLROnPlateau reducing learning rate to 0.0001.\n",
      "67/67 [==============================] - 155s 2s/step - loss: 0.2672 - categorical_accuracy: 0.9005 - val_loss: 0.7701 - val_categorical_accuracy: 0.7300 - lr: 1.0000e-04\n",
      "Epoch 20/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2837 - categorical_accuracy: 0.8944\n",
      "Epoch 00020: val_loss did not improve from 0.64960\n",
      "67/67 [==============================] - 154s 2s/step - loss: 0.2837 - categorical_accuracy: 0.8944 - val_loss: 0.6917 - val_categorical_accuracy: 0.7900 - lr: 1.0000e-04\n",
      "Epoch 21/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1768 - categorical_accuracy: 0.9502\n",
      "Epoch 00021: val_loss did not improve from 0.64960\n",
      "67/67 [==============================] - 156s 2s/step - loss: 0.1768 - categorical_accuracy: 0.9502 - val_loss: 0.7031 - val_categorical_accuracy: 0.7700 - lr: 1.0000e-04\n",
      "Epoch 22/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1843 - categorical_accuracy: 0.9397\n",
      "Epoch 00022: val_loss improved from 0.64960 to 0.64278, saving model to model_conv3D_6_2023-08-2816_51_10.362060/model-00022-0.18432-0.93967-0.64278-0.78000.h5\n",
      "67/67 [==============================] - 156s 2s/step - loss: 0.1843 - categorical_accuracy: 0.9397 - val_loss: 0.6428 - val_categorical_accuracy: 0.7800 - lr: 1.0000e-04\n",
      "Epoch 23/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2432 - categorical_accuracy: 0.9276\n",
      "Epoch 00023: val_loss did not improve from 0.64278\n",
      "67/67 [==============================] - 152s 2s/step - loss: 0.2432 - categorical_accuracy: 0.9276 - val_loss: 0.8899 - val_categorical_accuracy: 0.7200 - lr: 1.0000e-04\n",
      "Epoch 24/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1779 - categorical_accuracy: 0.9502\n",
      "Epoch 00024: val_loss improved from 0.64278 to 0.62921, saving model to model_conv3D_6_2023-08-2816_51_10.362060/model-00024-0.17787-0.95023-0.62921-0.81000.h5\n",
      "67/67 [==============================] - 154s 2s/step - loss: 0.1779 - categorical_accuracy: 0.9502 - val_loss: 0.6292 - val_categorical_accuracy: 0.8100 - lr: 1.0000e-04\n",
      "Epoch 25/25\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1971 - categorical_accuracy: 0.9397\n",
      "Epoch 00025: val_loss did not improve from 0.62921\n",
      "67/67 [==============================] - 154s 2s/step - loss: 0.1971 - categorical_accuracy: 0.9397 - val_loss: 0.8440 - val_categorical_accuracy: 0.7800 - lr: 1.0000e-04\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(train_generator, steps_per_epoch=steps_per_epoch, epochs=num_epochs, verbose=1, \n",
    "                              callbacks=callbacks_list, validation_data=val_generator, \n",
    "                              validation_steps=validation_steps, class_weight=None, workers=1, initial_epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "KiFYXDE3Nr_9"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['loss', 'categorical_accuracy', 'val_loss', 'val_categorical_accuracy', 'lr'])\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_loss_accuracy(history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "FNtKJ11nCSCy"
   },
   "source": [
    "#### Model 6: Conv 3D Model with 20 frames per video (32, 64, 128, 256) filters conv 3D layers + 128 dense layer + 128 dense layer + image size 120 by 120)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "6l8TrudWB3ao"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# epochs = 25\n",
      "# batch size = 20\n",
      "# num_frames per video = 20\n",
      "Model: \"sequential_2\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv3d_8 (Conv3D)           (None, 20, 120, 120, 32)  2624      \n",
      "                                                                 \n",
      " activation_14 (Activation)  (None, 20, 120, 120, 32)  0         \n",
      "                                                                 \n",
      " batch_normalization_12 (Bat  (None, 20, 120, 120, 32)  128      \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " max_pooling3d_8 (MaxPooling  (None, 10, 60, 60, 32)   0         \n",
      " 3D)                                                             \n",
      "                                                                 \n",
      " conv3d_9 (Conv3D)           (None, 10, 60, 60, 64)    55360     \n",
      "                                                                 \n",
      " activation_15 (Activation)  (None, 10, 60, 60, 64)    0         \n",
      "                                                                 \n",
      " batch_normalization_13 (Bat  (None, 10, 60, 60, 64)   256       \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " max_pooling3d_9 (MaxPooling  (None, 5, 30, 30, 64)    0         \n",
      " 3D)                                                             \n",
      "                                                                 \n",
      " conv3d_10 (Conv3D)          (None, 5, 30, 30, 128)    221312    \n",
      "                                                                 \n",
      " activation_16 (Activation)  (None, 5, 30, 30, 128)    0         \n",
      "                                                                 \n",
      " batch_normalization_14 (Bat  (None, 5, 30, 30, 128)   512       \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " max_pooling3d_10 (MaxPoolin  (None, 2, 15, 15, 128)   0         \n",
      " g3D)                                                            \n",
      "                                                                 \n",
      " conv3d_11 (Conv3D)          (None, 2, 15, 15, 256)    884992    \n",
      "                                                                 \n",
      " activation_17 (Activation)  (None, 2, 15, 15, 256)    0         \n",
      "                                                                 \n",
      " batch_normalization_15 (Bat  (None, 2, 15, 15, 256)   1024      \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " max_pooling3d_11 (MaxPoolin  (None, 1, 7, 7, 256)     0         \n",
      " g3D)                                                            \n",
      "                                                                 \n",
      " flatten_2 (Flatten)         (None, 12544)             0         \n",
      "                                                                 \n",
      " dense_6 (Dense)             (None, 128)               1605760   \n",
      "                                                                 \n",
      " activation_18 (Activation)  (None, 128)               0         \n",
      "                                                                 \n",
      " batch_normalization_16 (Bat  (None, 128)              512       \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " dropout_4 (Dropout)         (None, 128)               0         \n",
      "                                                                 \n",
      " dense_7 (Dense)             (None, 128)               16512     \n",
      "                                                                 \n",
      " activation_19 (Activation)  (None, 128)               0         \n",
      "                                                                 \n",
      " batch_normalization_17 (Bat  (None, 128)              512       \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " dropout_5 (Dropout)         (None, 128)               0         \n",
      "                                                                 \n",
      " dense_8 (Dense)             (None, 5)                 645       \n",
      "                                                                 \n",
      " activation_20 (Activation)  (None, 5)                 0         \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 2,790,149\n",
      "Trainable params: 2,788,677\n",
      "Non-trainable params: 1,472\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "num_epochs = 25 # choose the number of epochs\n",
    "print ('# epochs =', num_epochs)\n",
    "batch_size = 20  #experiment with the batch size\n",
    "print ('# batch size =', batch_size)\n",
    "num_frames = 20\n",
    "print ('# num_frames per video =', num_frames)\n",
    "height = 120\n",
    "width = 120\n",
    "\n",
    "conv_filters = (32, 64, 128, 256)\n",
    "dense_filters = (128, 128)\n",
    "model = conv3D(conv_filters=conv_filters,dense_nodes=dense_filters, num_images=num_frames)\n",
    "print (model.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "VXcwI6cvB3as"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:`period` argument is deprecated. Please use `save_freq` to specify the frequency in number of batches seen.\n"
     ]
    }
   ],
   "source": [
    "callbacks_list = model_callbacks(\"model_conv3D_6\")\n",
    "\n",
    "steps_per_epoch, validation_steps = calculate_steps(num_train_sequences, num_val_sequences, batch_size)\n",
    "\n",
    "train_generator = generate(train_path, train_doc, batch_size, augment=False, num_images=num_frames)\n",
    "\n",
    "val_generator   = generate(val_path, val_doc, batch_size, num_images=num_frames)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "x6ZEXx3YB3av"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/25\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-08-29 07:29:47.546507: I tensorflow/stream_executor/cuda/cuda_dnn.cc:377] Loaded cuDNN version 8302\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "34/34 [==============================] - ETA: 0s - loss: 1.5383 - categorical_accuracy: 0.4284\n",
      "Epoch 00001: val_loss improved from inf to 1.46857, saving model to model_conv3D_8_2023-08-2907_28_23.340432/model-00001-1.53832-0.42836-1.46857-0.36000.h5\n",
      "34/34 [==============================] - 112s 3s/step - loss: 1.5383 - categorical_accuracy: 0.4284 - val_loss: 1.4686 - val_categorical_accuracy: 0.3600 - lr: 0.0010\n",
      "Epoch 2/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 1.0426 - categorical_accuracy: 0.6078\n",
      "Epoch 00002: val_loss did not improve from 1.46857\n",
      "34/34 [==============================] - 110s 3s/step - loss: 1.0426 - categorical_accuracy: 0.6078 - val_loss: 5.5039 - val_categorical_accuracy: 0.1500 - lr: 0.0010\n",
      "Epoch 3/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.8064 - categorical_accuracy: 0.6938\n",
      "Epoch 00003: val_loss did not improve from 1.46857\n",
      "34/34 [==============================] - 109s 3s/step - loss: 0.8064 - categorical_accuracy: 0.6938 - val_loss: 5.6381 - val_categorical_accuracy: 0.2100 - lr: 0.0010\n",
      "Epoch 4/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.6032 - categorical_accuracy: 0.7919\n",
      "Epoch 00004: val_loss did not improve from 1.46857\n",
      "\n",
      "Epoch 00004: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.\n",
      "34/34 [==============================] - 110s 3s/step - loss: 0.6032 - categorical_accuracy: 0.7919 - val_loss: 6.6506 - val_categorical_accuracy: 0.2000 - lr: 0.0010\n",
      "Epoch 5/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.4827 - categorical_accuracy: 0.8235\n",
      "Epoch 00005: val_loss did not improve from 1.46857\n",
      "34/34 [==============================] - 109s 3s/step - loss: 0.4827 - categorical_accuracy: 0.8235 - val_loss: 6.2039 - val_categorical_accuracy: 0.2000 - lr: 1.0000e-04\n",
      "Epoch 6/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.4380 - categorical_accuracy: 0.8507\n",
      "Epoch 00006: val_loss did not improve from 1.46857\n",
      "34/34 [==============================] - 109s 3s/step - loss: 0.4380 - categorical_accuracy: 0.8507 - val_loss: 6.6328 - val_categorical_accuracy: 0.2300 - lr: 1.0000e-04\n",
      "Epoch 7/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.3610 - categorical_accuracy: 0.8582\n",
      "Epoch 00007: val_loss did not improve from 1.46857\n",
      "\n",
      "Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.0001.\n",
      "34/34 [==============================] - 109s 3s/step - loss: 0.3610 - categorical_accuracy: 0.8582 - val_loss: 5.6800 - val_categorical_accuracy: 0.2400 - lr: 1.0000e-04\n",
      "Epoch 8/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.3477 - categorical_accuracy: 0.8763\n",
      "Epoch 00008: val_loss did not improve from 1.46857\n",
      "34/34 [==============================] - 110s 3s/step - loss: 0.3477 - categorical_accuracy: 0.8763 - val_loss: 5.2005 - val_categorical_accuracy: 0.1500 - lr: 1.0000e-04\n",
      "Epoch 9/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.3049 - categorical_accuracy: 0.8899\n",
      "Epoch 00009: val_loss did not improve from 1.46857\n",
      "34/34 [==============================] - 108s 3s/step - loss: 0.3049 - categorical_accuracy: 0.8899 - val_loss: 3.8534 - val_categorical_accuracy: 0.3200 - lr: 1.0000e-04\n",
      "Epoch 10/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.2870 - categorical_accuracy: 0.8989\n",
      "Epoch 00010: val_loss did not improve from 1.46857\n",
      "34/34 [==============================] - 110s 3s/step - loss: 0.2870 - categorical_accuracy: 0.8989 - val_loss: 4.5881 - val_categorical_accuracy: 0.2200 - lr: 1.0000e-04\n",
      "Epoch 11/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.3101 - categorical_accuracy: 0.8869\n",
      "Epoch 00011: val_loss did not improve from 1.46857\n",
      "34/34 [==============================] - 110s 3s/step - loss: 0.3101 - categorical_accuracy: 0.8869 - val_loss: 3.4011 - val_categorical_accuracy: 0.3200 - lr: 1.0000e-04\n",
      "Epoch 12/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.2494 - categorical_accuracy: 0.9216\n",
      "Epoch 00012: val_loss did not improve from 1.46857\n",
      "34/34 [==============================] - 110s 3s/step - loss: 0.2494 - categorical_accuracy: 0.9216 - val_loss: 3.0363 - val_categorical_accuracy: 0.3800 - lr: 1.0000e-04\n",
      "Epoch 13/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.2063 - categorical_accuracy: 0.9321\n",
      "Epoch 00013: val_loss did not improve from 1.46857\n",
      "34/34 [==============================] - 110s 3s/step - loss: 0.2063 - categorical_accuracy: 0.9321 - val_loss: 2.6609 - val_categorical_accuracy: 0.4300 - lr: 1.0000e-04\n",
      "Epoch 14/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.1892 - categorical_accuracy: 0.9487\n",
      "Epoch 00014: val_loss did not improve from 1.46857\n",
      "34/34 [==============================] - 109s 3s/step - loss: 0.1892 - categorical_accuracy: 0.9487 - val_loss: 1.8616 - val_categorical_accuracy: 0.5200 - lr: 1.0000e-04\n",
      "Epoch 15/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.1844 - categorical_accuracy: 0.9442\n",
      "Epoch 00015: val_loss did not improve from 1.46857\n",
      "34/34 [==============================] - 109s 3s/step - loss: 0.1844 - categorical_accuracy: 0.9442 - val_loss: 1.5264 - val_categorical_accuracy: 0.5400 - lr: 1.0000e-04\n",
      "Epoch 16/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.2415 - categorical_accuracy: 0.9261\n",
      "Epoch 00016: val_loss improved from 1.46857 to 0.96045, saving model to model_conv3D_8_2023-08-2907_28_23.340432/model-00016-0.24150-0.92609-0.96045-0.70000.h5\n",
      "34/34 [==============================] - 109s 3s/step - loss: 0.2415 - categorical_accuracy: 0.9261 - val_loss: 0.9604 - val_categorical_accuracy: 0.7000 - lr: 1.0000e-04\n",
      "Epoch 17/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.1699 - categorical_accuracy: 0.9563\n",
      "Epoch 00017: val_loss did not improve from 0.96045\n",
      "34/34 [==============================] - 109s 3s/step - loss: 0.1699 - categorical_accuracy: 0.9563 - val_loss: 1.0628 - val_categorical_accuracy: 0.6700 - lr: 1.0000e-04\n",
      "Epoch 18/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.1605 - categorical_accuracy: 0.9487\n",
      "Epoch 00018: val_loss improved from 0.96045 to 0.82900, saving model to model_conv3D_8_2023-08-2907_28_23.340432/model-00018-0.16052-0.94872-0.82900-0.73000.h5\n",
      "34/34 [==============================] - 109s 3s/step - loss: 0.1605 - categorical_accuracy: 0.9487 - val_loss: 0.8290 - val_categorical_accuracy: 0.7300 - lr: 1.0000e-04\n",
      "Epoch 19/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.1490 - categorical_accuracy: 0.9653\n",
      "Epoch 00019: val_loss did not improve from 0.82900\n",
      "34/34 [==============================] - 108s 3s/step - loss: 0.1490 - categorical_accuracy: 0.9653 - val_loss: 0.9152 - val_categorical_accuracy: 0.6300 - lr: 1.0000e-04\n",
      "Epoch 20/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.1280 - categorical_accuracy: 0.9713\n",
      "Epoch 00020: val_loss improved from 0.82900 to 0.55831, saving model to model_conv3D_8_2023-08-2907_28_23.340432/model-00020-0.12803-0.97134-0.55831-0.75000.h5\n",
      "34/34 [==============================] - 110s 3s/step - loss: 0.1280 - categorical_accuracy: 0.9713 - val_loss: 0.5583 - val_categorical_accuracy: 0.7500 - lr: 1.0000e-04\n",
      "Epoch 21/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.1291 - categorical_accuracy: 0.9713\n",
      "Epoch 00021: val_loss did not improve from 0.55831\n",
      "34/34 [==============================] - 109s 3s/step - loss: 0.1291 - categorical_accuracy: 0.9713 - val_loss: 0.6309 - val_categorical_accuracy: 0.7600 - lr: 1.0000e-04\n",
      "Epoch 22/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.1314 - categorical_accuracy: 0.9593\n",
      "Epoch 00022: val_loss did not improve from 0.55831\n",
      "34/34 [==============================] - 106s 3s/step - loss: 0.1314 - categorical_accuracy: 0.9593 - val_loss: 0.7507 - val_categorical_accuracy: 0.7300 - lr: 1.0000e-04\n",
      "Epoch 23/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.1230 - categorical_accuracy: 0.9774\n",
      "Epoch 00023: val_loss improved from 0.55831 to 0.50537, saving model to model_conv3D_8_2023-08-2907_28_23.340432/model-00023-0.12296-0.97738-0.50537-0.79000.h5\n",
      "34/34 [==============================] - 110s 3s/step - loss: 0.1230 - categorical_accuracy: 0.9774 - val_loss: 0.5054 - val_categorical_accuracy: 0.7900 - lr: 1.0000e-04\n",
      "Epoch 24/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.1026 - categorical_accuracy: 0.9729\n",
      "Epoch 00024: val_loss improved from 0.50537 to 0.48592, saving model to model_conv3D_8_2023-08-2907_28_23.340432/model-00024-0.10259-0.97285-0.48592-0.85000.h5\n",
      "34/34 [==============================] - 109s 3s/step - loss: 0.1026 - categorical_accuracy: 0.9729 - val_loss: 0.4859 - val_categorical_accuracy: 0.8500 - lr: 1.0000e-04\n",
      "Epoch 25/25\n",
      "34/34 [==============================] - ETA: 0s - loss: 0.0857 - categorical_accuracy: 0.9834\n",
      "Epoch 00025: val_loss improved from 0.48592 to 0.44718, saving model to model_conv3D_8_2023-08-2907_28_23.340432/model-00025-0.08569-0.98341-0.44718-0.84000.h5\n",
      "34/34 [==============================] - 111s 3s/step - loss: 0.0857 - categorical_accuracy: 0.9834 - val_loss: 0.4472 - val_categorical_accuracy: 0.8400 - lr: 1.0000e-04\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(train_generator, steps_per_epoch=steps_per_epoch, epochs=num_epochs, verbose=1, \n",
    "                              callbacks=callbacks_list, validation_data=val_generator, \n",
    "                              validation_steps=validation_steps, class_weight=None, workers=1, initial_epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "zoGer63EB3ax"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['loss', 'categorical_accuracy', 'val_loss', 'val_categorical_accuracy', 'lr'])\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_loss_accuracy(history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "q6YqsPFpy1ss"
   },
   "source": [
    "#### Model 7: Conv 3D Model with 30 frames per video and a 5 by 5 filter size (16, 32, 64, 128 filters conv 3D layers + 256 dense layer + 128 dense layer + image size 120 by 120)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "MGb6a5vey1sg"
   },
   "outputs": [],
   "source": [
    "#write your model here\n",
    "def conv3D_5by5(conv_filters=(16, 32, 64, 128), dense_nodes=(256,128), dropout=0.25, num_images=20, height=120, width=120):\n",
    "\n",
    "    model = Sequential()\n",
    "\n",
    "    model.add(Conv3D(conv_filters[0], (5, 5, 5), padding='same', input_shape=(num_images, height, width, 3)))\n",
    "    model.add(Activation('relu'))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(MaxPooling3D(pool_size=(2, 2, 2)))\n",
    "\n",
    "    model.add(Conv3D(conv_filters[1], (3, 3, 3), padding='same'))\n",
    "    model.add(Activation('relu'))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(MaxPooling3D(pool_size=(2, 2, 2)))\n",
    "    \n",
    "    model.add(Conv3D(conv_filters[2], (3, 3, 3), padding='same'))\n",
    "    model.add(Activation('relu'))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(MaxPooling3D(pool_size=(2, 2, 2)))\n",
    "    \n",
    "    model.add(Conv3D(conv_filters[3], (3, 3, 3), padding='same'))    \n",
    "    model.add(Activation('relu'))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(MaxPooling3D(pool_size=(2, 2, 2)))\n",
    "    \n",
    "    model.add(Flatten())\n",
    "    model.add(Dense(dense_nodes[0]))\n",
    "    model.add(Activation('relu'))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(Dropout(dropout))\n",
    "\n",
    "    model.add(Dense(dense_nodes[1]))\n",
    "    model.add(Activation('relu'))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(Dropout(dropout))\n",
    "\n",
    "    model.add(Dense(num_classes))\n",
    "    model.add(Activation('softmax'))\n",
    "\n",
    "    opt = tf.keras.optimizers.Adam() #write your optimizer\n",
    "    model.compile(opt, loss='categorical_crossentropy', metrics=['categorical_accuracy'])    \n",
    "    \n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "xekrJwF2y1ss"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# epochs = 15\n",
      "# batch size = 5\n",
      "# num_frames per video = 30\n",
      "Model: \"sequential_5\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv3d_20 (Conv3D)          (None, 30, 120, 120, 16)  6016      \n",
      "                                                                 \n",
      " activation_35 (Activation)  (None, 30, 120, 120, 16)  0         \n",
      "                                                                 \n",
      " batch_normalization_30 (Bat  (None, 30, 120, 120, 16)  64       \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " max_pooling3d_20 (MaxPoolin  (None, 15, 60, 60, 16)   0         \n",
      " g3D)                                                            \n",
      "                                                                 \n",
      " conv3d_21 (Conv3D)          (None, 15, 60, 60, 32)    13856     \n",
      "                                                                 \n",
      " activation_36 (Activation)  (None, 15, 60, 60, 32)    0         \n",
      "                                                                 \n",
      " batch_normalization_31 (Bat  (None, 15, 60, 60, 32)   128       \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " max_pooling3d_21 (MaxPoolin  (None, 7, 30, 30, 32)    0         \n",
      " g3D)                                                            \n",
      "                                                                 \n",
      " conv3d_22 (Conv3D)          (None, 7, 30, 30, 64)     55360     \n",
      "                                                                 \n",
      " activation_37 (Activation)  (None, 7, 30, 30, 64)     0         \n",
      "                                                                 \n",
      " batch_normalization_32 (Bat  (None, 7, 30, 30, 64)    256       \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " max_pooling3d_22 (MaxPoolin  (None, 3, 15, 15, 64)    0         \n",
      " g3D)                                                            \n",
      "                                                                 \n",
      " conv3d_23 (Conv3D)          (None, 3, 15, 15, 128)    221312    \n",
      "                                                                 \n",
      " activation_38 (Activation)  (None, 3, 15, 15, 128)    0         \n",
      "                                                                 \n",
      " batch_normalization_33 (Bat  (None, 3, 15, 15, 128)   512       \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " max_pooling3d_23 (MaxPoolin  (None, 1, 7, 7, 128)     0         \n",
      " g3D)                                                            \n",
      "                                                                 \n",
      " flatten_5 (Flatten)         (None, 6272)              0         \n",
      "                                                                 \n",
      " dense_15 (Dense)            (None, 256)               1605888   \n",
      "                                                                 \n",
      " activation_39 (Activation)  (None, 256)               0         \n",
      "                                                                 \n",
      " batch_normalization_34 (Bat  (None, 256)              1024      \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " dropout_10 (Dropout)        (None, 256)               0         \n",
      "                                                                 \n",
      " dense_16 (Dense)            (None, 128)               32896     \n",
      "                                                                 \n",
      " activation_40 (Activation)  (None, 128)               0         \n",
      "                                                                 \n",
      " batch_normalization_35 (Bat  (None, 128)              512       \n",
      " chNormalization)                                                \n",
      "                                                                 \n",
      " dropout_11 (Dropout)        (None, 128)               0         \n",
      "                                                                 \n",
      " dense_17 (Dense)            (None, 5)                 645       \n",
      "                                                                 \n",
      " activation_41 (Activation)  (None, 5)                 0         \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 1,938,469\n",
      "Trainable params: 1,937,221\n",
      "Non-trainable params: 1,248\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "num_epochs = 15 # choose the number of epochs\n",
    "print ('# epochs =', num_epochs)\n",
    "batch_size = 5  #experiment with the batch size\n",
    "print ('# batch size =', batch_size)\n",
    "num_frames = 30\n",
    "print ('# num_frames per video =', num_frames)\n",
    "height = 120\n",
    "width = 120\n",
    "\n",
    "model = conv3D_5by5(num_images=num_frames)\n",
    "\n",
    "print (model.summary())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Ng0cjVwUy1sx"
   },
   "source": [
    "The `steps_per_epoch` and `validation_steps` are used by `fit_generator` to decide the number of next() calls it need to make.\n",
    "Let us now fit the model. This will start training the model and with the help of the checkpoints, you'll be able to save the model at the end of each epoch."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "3JjMNoSBy1sx"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:`period` argument is deprecated. Please use `save_freq` to specify the frequency in number of batches seen.\n"
     ]
    }
   ],
   "source": [
    "callbacks_list = model_callbacks(\"model_conv3D_9\")\n",
    "\n",
    "steps_per_epoch, validation_steps = calculate_steps(num_train_sequences, num_val_sequences, batch_size)\n",
    "\n",
    "train_generator = generate(train_path, train_doc, batch_size, num_images=num_frames)\n",
    "\n",
    "val_generator   = generate(val_path, val_doc, batch_size, num_images=num_frames)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Oezy0OJly1s0"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/15\n",
      "133/133 [==============================] - ETA: 0s - loss: 1.7426 - categorical_accuracy: 0.3575\n",
      "Epoch 00001: val_loss improved from inf to 9.84336, saving model to model_conv3D_9_2023-08-2907_28_23.340432/model-00001-1.74264-0.35747-9.84336-0.15000.h5\n",
      "133/133 [==============================] - 166s 1s/step - loss: 1.7426 - categorical_accuracy: 0.3575 - val_loss: 9.8434 - val_categorical_accuracy: 0.1500 - lr: 0.0010\n",
      "Epoch 2/15\n",
      "133/133 [==============================] - ETA: 0s - loss: 1.4055 - categorical_accuracy: 0.4449\n",
      "Epoch 00002: val_loss improved from 9.84336 to 3.31989, saving model to model_conv3D_9_2023-08-2907_28_23.340432/model-00002-1.40553-0.44495-3.31989-0.27000.h5\n",
      "133/133 [==============================] - 164s 1s/step - loss: 1.4055 - categorical_accuracy: 0.4449 - val_loss: 3.3199 - val_categorical_accuracy: 0.2700 - lr: 0.0010\n",
      "Epoch 3/15\n",
      "133/133 [==============================] - ETA: 0s - loss: 1.3764 - categorical_accuracy: 0.4615\n",
      "Epoch 00003: val_loss improved from 3.31989 to 1.45191, saving model to model_conv3D_9_2023-08-2907_28_23.340432/model-00003-1.37643-0.46154-1.45191-0.45000.h5\n",
      "133/133 [==============================] - 163s 1s/step - loss: 1.3764 - categorical_accuracy: 0.4615 - val_loss: 1.4519 - val_categorical_accuracy: 0.4500 - lr: 0.0010\n",
      "Epoch 4/15\n",
      "133/133 [==============================] - ETA: 0s - loss: 1.2662 - categorical_accuracy: 0.5098\n",
      "Epoch 00004: val_loss did not improve from 1.45191\n",
      "133/133 [==============================] - 162s 1s/step - loss: 1.2662 - categorical_accuracy: 0.5098 - val_loss: 3.5605 - val_categorical_accuracy: 0.1800 - lr: 0.0010\n",
      "Epoch 5/15\n",
      "133/133 [==============================] - ETA: 0s - loss: 1.1678 - categorical_accuracy: 0.5686\n",
      "Epoch 00005: val_loss did not improve from 1.45191\n",
      "133/133 [==============================] - 163s 1s/step - loss: 1.1678 - categorical_accuracy: 0.5686 - val_loss: 4.8551 - val_categorical_accuracy: 0.2500 - lr: 0.0010\n",
      "Epoch 6/15\n",
      "133/133 [==============================] - ETA: 0s - loss: 1.0812 - categorical_accuracy: 0.5958\n",
      "Epoch 00006: val_loss did not improve from 1.45191\n",
      "\n",
      "Epoch 00006: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.\n",
      "133/133 [==============================] - 167s 1s/step - loss: 1.0812 - categorical_accuracy: 0.5958 - val_loss: 1.9101 - val_categorical_accuracy: 0.5100 - lr: 0.0010\n",
      "Epoch 7/15\n",
      "133/133 [==============================] - ETA: 0s - loss: 0.8636 - categorical_accuracy: 0.6727\n",
      "Epoch 00007: val_loss improved from 1.45191 to 0.96665, saving model to model_conv3D_9_2023-08-2907_28_23.340432/model-00007-0.86359-0.67270-0.96665-0.62000.h5\n",
      "133/133 [==============================] - 168s 1s/step - loss: 0.8636 - categorical_accuracy: 0.6727 - val_loss: 0.9667 - val_categorical_accuracy: 0.6200 - lr: 1.0000e-04\n",
      "Epoch 8/15\n",
      "133/133 [==============================] - ETA: 0s - loss: 0.8175 - categorical_accuracy: 0.6621\n",
      "Epoch 00008: val_loss improved from 0.96665 to 0.59153, saving model to model_conv3D_9_2023-08-2907_28_23.340432/model-00008-0.81748-0.66214-0.59153-0.81000.h5\n",
      "133/133 [==============================] - 168s 1s/step - loss: 0.8175 - categorical_accuracy: 0.6621 - val_loss: 0.5915 - val_categorical_accuracy: 0.8100 - lr: 1.0000e-04\n",
      "Epoch 9/15\n",
      "133/133 [==============================] - ETA: 0s - loss: 0.8442 - categorical_accuracy: 0.6742\n",
      "Epoch 00009: val_loss did not improve from 0.59153\n",
      "133/133 [==============================] - 171s 1s/step - loss: 0.8442 - categorical_accuracy: 0.6742 - val_loss: 0.6565 - val_categorical_accuracy: 0.7400 - lr: 1.0000e-04\n",
      "Epoch 10/15\n",
      "133/133 [==============================] - ETA: 0s - loss: 0.7705 - categorical_accuracy: 0.6968\n",
      "Epoch 00010: val_loss did not improve from 0.59153\n",
      "133/133 [==============================] - 162s 1s/step - loss: 0.7705 - categorical_accuracy: 0.6968 - val_loss: 0.6175 - val_categorical_accuracy: 0.8000 - lr: 1.0000e-04\n",
      "Epoch 11/15\n",
      "133/133 [==============================] - ETA: 0s - loss: 0.7539 - categorical_accuracy: 0.7074\n",
      "Epoch 00011: val_loss did not improve from 0.59153\n",
      "\n",
      "Epoch 00011: ReduceLROnPlateau reducing learning rate to 0.0001.\n",
      "133/133 [==============================] - 160s 1s/step - loss: 0.7539 - categorical_accuracy: 0.7074 - val_loss: 0.5951 - val_categorical_accuracy: 0.7500 - lr: 1.0000e-04\n",
      "Epoch 12/15\n",
      "133/133 [==============================] - ETA: 0s - loss: 0.7216 - categorical_accuracy: 0.7270\n",
      "Epoch 00012: val_loss improved from 0.59153 to 0.44703, saving model to model_conv3D_9_2023-08-2907_28_23.340432/model-00012-0.72157-0.72700-0.44703-0.86000.h5\n",
      "133/133 [==============================] - 159s 1s/step - loss: 0.7216 - categorical_accuracy: 0.7270 - val_loss: 0.4470 - val_categorical_accuracy: 0.8600 - lr: 1.0000e-04\n",
      "Epoch 13/15\n",
      "133/133 [==============================] - ETA: 0s - loss: 0.7623 - categorical_accuracy: 0.7149\n",
      "Epoch 00013: val_loss did not improve from 0.44703\n",
      "133/133 [==============================] - 162s 1s/step - loss: 0.7623 - categorical_accuracy: 0.7149 - val_loss: 0.4496 - val_categorical_accuracy: 0.8700 - lr: 1.0000e-04\n",
      "Epoch 14/15\n",
      "133/133 [==============================] - ETA: 0s - loss: 0.6455 - categorical_accuracy: 0.7496\n",
      "Epoch 00014: val_loss did not improve from 0.44703\n",
      "133/133 [==============================] - 160s 1s/step - loss: 0.6455 - categorical_accuracy: 0.7496 - val_loss: 1.1574 - val_categorical_accuracy: 0.6500 - lr: 1.0000e-04\n",
      "Epoch 15/15\n",
      "133/133 [==============================] - ETA: 0s - loss: 0.6610 - categorical_accuracy: 0.7466\n",
      "Epoch 00015: val_loss did not improve from 0.44703\n",
      "133/133 [==============================] - 158s 1s/step - loss: 0.6610 - categorical_accuracy: 0.7466 - val_loss: 0.4644 - val_categorical_accuracy: 0.8200 - lr: 1.0000e-04\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(train_generator, steps_per_epoch=steps_per_epoch, epochs=num_epochs, verbose=1, \n",
    "                              callbacks=callbacks_list, validation_data=val_generator, \n",
    "                              validation_steps=validation_steps, class_weight=None, workers=1, initial_epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Cb4VZNdGy1s2"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['loss', 'categorical_accuracy', 'val_loss', 'val_categorical_accuracy', 'lr'])\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_loss_accuracy(history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "3fFqKH6PCjEZ"
   },
   "source": [
    "### TRANSFER LEARNING - MobileNet Conv2d + GRU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "zmjMTdCtCjEd"
   },
   "outputs": [],
   "source": [
    "from keras.applications import mobilenet\n",
    "\n",
    "def mobilenet_RNN(fine_tune=False, gru=False, lstm_gru_cells=128, dense_nodes=128, dropout=0.25, num_images=20, height=120, width=120):\n",
    "\n",
    "    mobilenet_transfer = mobilenet.MobileNet(weights='imagenet', include_top=False)\n",
    "    print(\"Number of layers in the pretrained mobilenet model: \", len(mobilenet_transfer.layers))\n",
    "\n",
    "    if (fine_tune):\n",
    "        # Fine-tune from this layer onwards\n",
    "        fine_tune_at = 50\n",
    "        # Freeze all the layers before the `fine_tune_at` layer\n",
    "        for layer in mobilenet_transfer.layers[:fine_tune_at]:\n",
    "            layer.trainable = False\n",
    "\n",
    "    model = Sequential()\n",
    "    model.add(TimeDistributed(mobilenet_transfer,input_shape=(num_images, height, width, 3)))\n",
    "    model.add(TimeDistributed(BatchNormalization()))\n",
    "    model.add(TimeDistributed(MaxPooling2D((2, 2))))\n",
    "    model.add(TimeDistributed(Flatten()))\n",
    "    if (gru):\n",
    "        model.add(GRU(lstm_gru_cells))\n",
    "    else:\n",
    "        model.add(LSTM(lstm_gru_cells))\n",
    "    model.add(Dropout(dropout))\n",
    "    model.add(Dense(dense_nodes, activation='relu'))\n",
    "    model.add(Dropout(dropout))\n",
    "    model.add(Dense(num_classes, activation='softmax'))\n",
    "\n",
    "    opt = tf.keras.optimizers.Adam() #write your optimizer\n",
    "    model.compile(opt, loss='categorical_crossentropy', metrics=['categorical_accuracy'])    \n",
    "    return model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "q57Ui9iQp_Xb"
   },
   "source": [
    "#### Model 8: Mobilenet (re train all weights) + GRU (128 cells) + 128 dense nodes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 709
    },
    "colab_type": "code",
    "id": "sY1lGmf8CjEl",
    "outputId": "777fe671-779a-4126-ef91-5503d3adba0b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# epochs = 20\n",
      "# batch size = 10\n",
      "# num frames per video = 20\n",
      "WARNING:tensorflow:`input_shape` is undefined or non-square, or `rows` is not in [128, 160, 192, 224]. Weights for input shape (224, 224) will be loaded as the default.\n",
      "Number of layers in the pretrained mobilenet model:  86\n",
      "Model: \"sequential_6\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " time_distributed (TimeDistr  (None, 20, 3, 3, 1024)   3228864   \n",
      " ibuted)                                                         \n",
      "                                                                 \n",
      " time_distributed_1 (TimeDis  (None, 20, 3, 3, 1024)   4096      \n",
      " tributed)                                                       \n",
      "                                                                 \n",
      " time_distributed_2 (TimeDis  (None, 20, 1, 1, 1024)   0         \n",
      " tributed)                                                       \n",
      "                                                                 \n",
      " time_distributed_3 (TimeDis  (None, 20, 1024)         0         \n",
      " tributed)                                                       \n",
      "                                                                 \n",
      " gru (GRU)                   (None, 128)               443136    \n",
      "                                                                 \n",
      " dropout_12 (Dropout)        (None, 128)               0         \n",
      "                                                                 \n",
      " dense_18 (Dense)            (None, 128)               16512     \n",
      "                                                                 \n",
      " dropout_13 (Dropout)        (None, 128)               0         \n",
      "                                                                 \n",
      " dense_19 (Dense)            (None, 5)                 645       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 3,693,253\n",
      "Trainable params: 3,669,317\n",
      "Non-trainable params: 23,936\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "num_epochs = 20 # choose the number of epochs\n",
    "print ('# epochs =', num_epochs)\n",
    "batch_size = 10  #experiment with the batch size\n",
    "print ('# batch size =', batch_size)\n",
    "num_frames = 20\n",
    "print ('# num frames per video =', num_frames)\n",
    "height = 120\n",
    "width = 120\n",
    "\n",
    "model = mobilenet_RNN(num_images=num_frames, gru=True)\n",
    "print (model.summary())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "S3mHBipiCjEl"
   },
   "source": [
    "The `steps_per_epoch` and `validation_steps` are used by `fit_generator` to decide the number of next() calls it need to make.\n",
    "Let us now fit the model. This will start training the model and with the help of the checkpoints, you'll be able to save the model at the end of each epoch."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "oEbMc5wECjEq"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:`period` argument is deprecated. Please use `save_freq` to specify the frequency in number of batches seen.\n"
     ]
    }
   ],
   "source": [
    "callbacks_list = model_callbacks(\"model_retrain_mobilenet_gru\")\n",
    "\n",
    "steps_per_epoch, validation_steps = calculate_steps(num_train_sequences, num_val_sequences, batch_size)\n",
    "\n",
    "train_generator = generate(train_path, train_doc, batch_size, num_images=num_frames)\n",
    "\n",
    "val_generator   = generate(val_path, val_doc, batch_size, num_images=num_frames)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 87
    },
    "colab_type": "code",
    "id": "gytABaPWPrIg",
    "outputId": "cb3d746b-1081-41bd-a98a-96c6e9503504"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.8684 - categorical_accuracy: 0.6817\n",
      "Epoch 00001: val_loss improved from inf to 0.73860, saving model to model_retrain_mobilenet_gru_2023-08-2907_28_23.340432/model-00001-0.86838-0.68175-0.73860-0.70000.h5\n",
      "67/67 [==============================] - 115s 2s/step - loss: 0.8684 - categorical_accuracy: 0.6817 - val_loss: 0.7386 - val_categorical_accuracy: 0.7000 - lr: 0.0010\n",
      "Epoch 2/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2497 - categorical_accuracy: 0.9261\n",
      "Epoch 00002: val_loss improved from 0.73860 to 0.27647, saving model to model_retrain_mobilenet_gru_2023-08-2907_28_23.340432/model-00002-0.24975-0.92609-0.27647-0.89000.h5\n",
      "67/67 [==============================] - 112s 2s/step - loss: 0.2497 - categorical_accuracy: 0.9261 - val_loss: 0.2765 - val_categorical_accuracy: 0.8900 - lr: 0.0010\n",
      "Epoch 3/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2176 - categorical_accuracy: 0.9397\n",
      "Epoch 00003: val_loss improved from 0.27647 to 0.20935, saving model to model_retrain_mobilenet_gru_2023-08-2907_28_23.340432/model-00003-0.21755-0.93967-0.20935-0.88000.h5\n",
      "67/67 [==============================] - 109s 2s/step - loss: 0.2176 - categorical_accuracy: 0.9397 - val_loss: 0.2094 - val_categorical_accuracy: 0.8800 - lr: 0.0010\n",
      "Epoch 4/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1917 - categorical_accuracy: 0.9532\n",
      "Epoch 00004: val_loss did not improve from 0.20935\n",
      "67/67 [==============================] - 109s 2s/step - loss: 0.1917 - categorical_accuracy: 0.9532 - val_loss: 0.2642 - val_categorical_accuracy: 0.9200 - lr: 0.0010\n",
      "Epoch 5/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1714 - categorical_accuracy: 0.9457\n",
      "Epoch 00005: val_loss did not improve from 0.20935\n",
      "67/67 [==============================] - 111s 2s/step - loss: 0.1714 - categorical_accuracy: 0.9457 - val_loss: 0.2690 - val_categorical_accuracy: 0.9000 - lr: 0.0010\n",
      "Epoch 6/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1799 - categorical_accuracy: 0.9532\n",
      "Epoch 00006: val_loss did not improve from 0.20935\n",
      "\n",
      "Epoch 00006: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.\n",
      "67/67 [==============================] - 111s 2s/step - loss: 0.1799 - categorical_accuracy: 0.9532 - val_loss: 0.3204 - val_categorical_accuracy: 0.9100 - lr: 0.0010\n",
      "Epoch 7/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0928 - categorical_accuracy: 0.9789\n",
      "Epoch 00007: val_loss improved from 0.20935 to 0.06929, saving model to model_retrain_mobilenet_gru_2023-08-2907_28_23.340432/model-00007-0.09277-0.97888-0.06929-0.98000.h5\n",
      "67/67 [==============================] - 109s 2s/step - loss: 0.0928 - categorical_accuracy: 0.9789 - val_loss: 0.0693 - val_categorical_accuracy: 0.9800 - lr: 1.0000e-04\n",
      "Epoch 8/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0519 - categorical_accuracy: 0.9834\n",
      "Epoch 00008: val_loss did not improve from 0.06929\n",
      "67/67 [==============================] - 108s 2s/step - loss: 0.0519 - categorical_accuracy: 0.9834 - val_loss: 0.0706 - val_categorical_accuracy: 0.9700 - lr: 1.0000e-04\n",
      "Epoch 9/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0252 - categorical_accuracy: 0.9925\n",
      "Epoch 00009: val_loss did not improve from 0.06929\n",
      "67/67 [==============================] - 108s 2s/step - loss: 0.0252 - categorical_accuracy: 0.9925 - val_loss: 0.0946 - val_categorical_accuracy: 0.9600 - lr: 1.0000e-04\n",
      "Epoch 10/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0274 - categorical_accuracy: 0.9925\n",
      "Epoch 00010: val_loss improved from 0.06929 to 0.04789, saving model to model_retrain_mobilenet_gru_2023-08-2907_28_23.340432/model-00010-0.02742-0.99246-0.04789-0.98000.h5\n",
      "67/67 [==============================] - 112s 2s/step - loss: 0.0274 - categorical_accuracy: 0.9925 - val_loss: 0.0479 - val_categorical_accuracy: 0.9800 - lr: 1.0000e-04\n",
      "Epoch 11/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0248 - categorical_accuracy: 0.9970\n",
      "Epoch 00011: val_loss did not improve from 0.04789\n",
      "67/67 [==============================] - 114s 2s/step - loss: 0.0248 - categorical_accuracy: 0.9970 - val_loss: 0.0558 - val_categorical_accuracy: 0.9800 - lr: 1.0000e-04\n",
      "Epoch 12/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0238 - categorical_accuracy: 0.9940\n",
      "Epoch 00012: val_loss did not improve from 0.04789\n",
      "67/67 [==============================] - 114s 2s/step - loss: 0.0238 - categorical_accuracy: 0.9940 - val_loss: 0.0763 - val_categorical_accuracy: 0.9700 - lr: 1.0000e-04\n",
      "Epoch 13/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0166 - categorical_accuracy: 0.9970\n",
      "Epoch 00013: val_loss did not improve from 0.04789\n",
      "\n",
      "Epoch 00013: ReduceLROnPlateau reducing learning rate to 0.0001.\n",
      "67/67 [==============================] - 115s 2s/step - loss: 0.0166 - categorical_accuracy: 0.9970 - val_loss: 0.0737 - val_categorical_accuracy: 0.9700 - lr: 1.0000e-04\n",
      "Epoch 14/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0197 - categorical_accuracy: 0.9970\n",
      "Epoch 00014: val_loss improved from 0.04789 to 0.02212, saving model to model_retrain_mobilenet_gru_2023-08-2907_28_23.340432/model-00014-0.01965-0.99698-0.02212-1.00000.h5\n",
      "67/67 [==============================] - 116s 2s/step - loss: 0.0197 - categorical_accuracy: 0.9970 - val_loss: 0.0221 - val_categorical_accuracy: 1.0000 - lr: 1.0000e-04\n",
      "Epoch 15/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0119 - categorical_accuracy: 0.9970\n",
      "Epoch 00015: val_loss did not improve from 0.02212\n",
      "67/67 [==============================] - 120s 2s/step - loss: 0.0119 - categorical_accuracy: 0.9970 - val_loss: 0.0509 - val_categorical_accuracy: 0.9900 - lr: 1.0000e-04\n",
      "Epoch 16/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0090 - categorical_accuracy: 0.9985\n",
      "Epoch 00016: val_loss did not improve from 0.02212\n",
      "67/67 [==============================] - 116s 2s/step - loss: 0.0090 - categorical_accuracy: 0.9985 - val_loss: 0.0784 - val_categorical_accuracy: 0.9800 - lr: 1.0000e-04\n",
      "Epoch 17/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0077 - categorical_accuracy: 0.9985\n",
      "Epoch 00017: val_loss improved from 0.02212 to 0.01333, saving model to model_retrain_mobilenet_gru_2023-08-2907_28_23.340432/model-00017-0.00771-0.99849-0.01333-1.00000.h5\n",
      "67/67 [==============================] - 120s 2s/step - loss: 0.0077 - categorical_accuracy: 0.9985 - val_loss: 0.0133 - val_categorical_accuracy: 1.0000 - lr: 1.0000e-04\n",
      "Epoch 18/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0079 - categorical_accuracy: 0.9970\n",
      "Epoch 00018: val_loss did not improve from 0.01333\n",
      "67/67 [==============================] - 115s 2s/step - loss: 0.0079 - categorical_accuracy: 0.9970 - val_loss: 0.0811 - val_categorical_accuracy: 0.9800 - lr: 1.0000e-04\n",
      "Epoch 19/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0079 - categorical_accuracy: 0.9985\n",
      "Epoch 00019: val_loss did not improve from 0.01333\n",
      "67/67 [==============================] - 113s 2s/step - loss: 0.0079 - categorical_accuracy: 0.9985 - val_loss: 0.0460 - val_categorical_accuracy: 0.9900 - lr: 1.0000e-04\n",
      "Epoch 20/20\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0178 - categorical_accuracy: 0.9970\n",
      "Epoch 00020: val_loss did not improve from 0.01333\n",
      "67/67 [==============================] - 117s 2s/step - loss: 0.0178 - categorical_accuracy: 0.9970 - val_loss: 0.0465 - val_categorical_accuracy: 0.9900 - lr: 1.0000e-04\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(train_generator, steps_per_epoch=steps_per_epoch, epochs=num_epochs, verbose=1, \n",
    "                              callbacks=callbacks_list, validation_data=val_generator, \n",
    "                              validation_steps=validation_steps, class_weight=None, workers=1, initial_epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "icQzD_q2CjEs"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['loss', 'categorical_accuracy', 'val_loss', 'val_categorical_accuracy', 'lr'])\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8/fFQqAAAACXBIWXMAAAsTAAALEwEAmpwYAAA6WklEQVR4nO3dd3zU9f3A8dc7O5AJSQghLGU7GCJui7hw4Ubc2lbaqq222tYO56+ttrVqrbuWuncdqCiCgta6AAVFZA9JICEQMoBcyHj//vh8A0e8hAvJ9y4k7+fjcY987zvfd0nufd/PFFXFGGOMaSwm2gEYY4xpnyxBGGOMCckShDHGmJAsQRhjjAnJEoQxxpiQLEEYY4wJyRKEMYCIPCYifwhz39UicpzfMRkTbZYgjDHGhGQJwpgORETioh2D6TgsQZi9hle080sR+VJEtorIv0Skh4i8JSKVIjJTRDKD9p8gIl+LSJmIzBaRoUHbRorI595xzwNJja51qojM9479SEQODDPGU0TkCxGpEJG1InJLo+1Heucr87Zf5q1PFpG/icgaESkXkQ+9dWNFpCDE+3Cct3yLiLwkIk+JSAVwmYiMEZGPvWusF5H7RCQh6Pj9RGSGiJSKSLGI/FZEckVkm4h0D9pvlIiUiEh8OK/ddDyWIMze5mzgeGAQcBrwFvBbIBv39/wzABEZBDwLXOttmwa8LiIJ3oflq8CTQDfgRe+8eMeOBKYAPwK6Aw8DU0UkMYz4tgKXABnAKcBPROQM77x9vXj/4cU0ApjvHXcncBBwuBfTr4D6MN+T04GXvGs+DdQBPweygMOAY4ErvRhSgZnA20AeMAB4V1WLgNnAxKDzXgw8p6o1YcZhOhhLEGZv8w9VLVbVQuC/wKeq+oWqBoBXgJHefucBb6rqDO8D7k4gGfcBfCgQD9yjqjWq+hIwJ+gak4GHVfVTVa1T1ceBau+4ZqnqbFX9SlXrVfVLXJL6nrf5AmCmqj7rXXeTqs4XkRjg+8A1qlroXfMjVa0O8z35WFVf9a5ZparzVPUTVa1V1dW4BNcQw6lAkar+TVUDqlqpqp962x4HLgIQkVjgfFwSNZ2UJQiztykOWq4K8TzFW84D1jRsUNV6YC3Qy9tWqLuOVLkmaLkvcJ1XRFMmImVAb++4ZonIISIyyyuaKQd+jPsmj3eOFSEOy8IVcYXaFo61jWIYJCJviEiRV+z0pzBiAHgNGCYi/XF3aeWq+tkexmQ6AEsQpqNah/ugB0BEBPfhWAisB3p56xr0CVpeC/xRVTOCHl1U9dkwrvsMMBXorarpwENAw3XWAvuGOGYjEGhi21agS9DriMUVTwVrPCTzg8BiYKCqpuGK4IJj2CdU4N5d2Au4u4iLsbuHTs8ShOmoXgBOEZFjvUrW63DFRB8BHwO1wM9EJF5EzgLGBB37T+DH3t2AiEhXr/I5NYzrpgKlqhoQkTG4YqUGTwPHichEEYkTke4iMsK7u5kC3CUieSISKyKHeXUeS4Ek7/rxwO+B3dWFpAIVwBYRGQL8JGjbG0BPEblWRBJFJFVEDgna/gRwGTABSxCdniUI0yGp6hLcN+F/4L6hnwacpqrbVXU7cBbug7AUV1/xctCxc4ErgPuAzcByb99wXAncJiKVwE24RNVw3m+Bk3HJqhRXQT3c23w98BWuLqQU+DMQo6rl3jkfxd39bAV2adUUwvW4xFSJS3bPB8VQiSs+Og0oApYBxwRt/x+ucvxzVQ0udjOdkNiEQcaYYCLyHvCMqj4a7VhMdFmCMMbsICIHAzNwdSiV0Y7HRJcVMRljABCRx3F9JK615GDA7iCMMcY0we4gjDHGhNRhBvbKysrSfv36RTsMY4zZq8ybN2+jqjbuWwN0oATRr18/5s6dG+0wjDFmryIiTTZntiImY4wxIVmCMMYYE5IlCGOMMSF1mDqIUGpqaigoKCAQCEQ7FN8lJSWRn59PfLzN7WKMaRsdOkEUFBSQmppKv3792HXgzo5FVdm0aRMFBQX0798/2uEYYzoI34qYRGSKiGwQkYVNbBcRuVdEloubQnJU0LZLRWSZ97h0T2MIBAJ07969QycHABGhe/funeJOyRgTOX7WQTwGjG9m+0nAQO8xGTeGPSLSDbgZOAQ3BPPNEjTPcEt19OTQoLO8TmNM5PiWIFT1A9ywxU05HXhCnU+ADBHpCZwIzFDVUlXdjBs4rLlEY4yJpiVvw9evQG24M6S2M0vfgZWzox1FuxTNOohe7DpVYoG3rqn13yEik3F3H/Tp0yfULlFXVlbGM888w5VXXtmi404++WSeeeYZMjIy/AnMmLaw+E147kJAIbkbHHgejLoYeuzXJqffXltPZaCGykCt96ihwvvZsG5LdQ01dS0fUy62vobjC+/n0BI3Zcec3pezaPDVpCQnkZoUR2pSPKlJcaR5P1OT4oiLbdl36tq6+h1xVuyIudHP6loAUhPjdrluatB1G5bjW3j91tqrK6lV9RHgEYDRo0e3y1EHy8rKeOCBB76TIGpra4mLa/rtnzZtmt+hmSiqCNSwpKiSxUWVLCuuJEaEXhnJ5GUkk5eRRK+MZLJSEomJaduiw5q6eorKA6wrq2JdeRXrygIUllUR2F7X4nPlVy3mqjXXUJw0mOnZlzO67G0O+OxR4j59kG+ThvBJxil8kXYs1bFdmz2PAoGaul0+NBuSQHVt/W7jSI6PJSGuZR+cObqJv3A3I1nCk5xCYn0VE9f+m+rVn3FNzdVsIr3JazX+0E5LikfRHXFvCUpeVTW7f18T42IQgUDN7l9rUnzMLgkkzYtlUI9Urj1uUIveg3BEM0EU4uYIbpDvrSsExjZaPztiUbWxG264gRUrVjBixAji4+NJSkoiMzOTxYsXs3TpUs444wzWrl1LIBDgmmuuYfLkycDOoUO2bNnCSSedxJFHHslHH31Er169eO2110hOTo7yKzPh2F5bz4qSLTuSwZKiCpYUVbKufGeDgtTEOOpV2droQzo+VuiZ7hJGXkbyjgTSMz1px3LXxJ3/wqpK2bYaCsuqXAIoq2JdeWCX5xsqq2k8gHNml3hSklr2UZBTX8LvAjdQKqn8VH5F6aZMnmAQ6UmXcULt+5xSPZOJRX/jtKL7mB17OG/EHceXMUOhibqyxDj3wZveJYH8bl28D774Jr9VN3yrT9mTb9WrPoCXboTt2+D0f3Px/mcBUDv3SY54+3o+TbuNNcc+SHH6AbvcuVQ2unNpuCNYV1YFsOMDOz8j+TtJJPhuIC0pnpSg19WQ3LbX1rOlOjhJ1oR1/fXlAerq/fl+7Otw3yLSD3hDVfcPse0U4GrcFIyHAPeq6hivknoe0NCq6XPgIFVtrj6D0aNHa+OxmL755huGDh0KwK2vf82idRWte0GNDMtL4+bTmr+VXr16NaeeeioLFy5k9uzZnHLKKSxcuHBHc9TS0lK6detGVVUVBx98MO+//z7du3ffJUEMGDCAuXPnMmLECCZOnMiECRO46KKLvnOt4NdrIqu+Xiksq2JxUSVLi3cmg5UlW6n1/nnjY4V9s1MYnJvK4NxUhuSmMjg3jbz0JAAqvA+bhkdhmfumv977pl9U8d0PgvTkeHqmJ1FTV8+6ssB3vrEmxMV4ySSJvPTkXRONty45IbZlLzZQAVNOhPJC+MF0yAnxN6cKhZ/DF0/AV/+B7ZXQfQCMvBiGnw+pPVp2zbagCv+7B969DboPhPOehOzBu+6z/kt44WL32k78E4y5osmk1lGIyDxVHR1qm293ECLyLO5OIEtECnAtk+IBVPUhYBouOSwHtgGXe9tKReT/cHPzAty2u+SwNxkzZswufRXuvfdeXnnlFQDWrl3LsmXL6N69+y7H9O/fnxEjRgBw0EEHsXr16kiF26GVbt3OtK/WM+2r9ZRu3b7H56lXpXBz1S53APmZyQzJTeX4YT0Y1COVIblp7JPdtdlvu+nJ8aQnxzO0Z1rI7bV19WyorPaSh0saDckkIS6GsYNzvASQ5BVVJdO9a0LbtnCrq4EXL4WNS+HCl0InB3AfqvkHuceJf4KvX4UvnoSZN7sP6EHjYdQlMOA4iI1AQUagHF69Eha/AfudBRP+AYkp392v54EweTa88mN465dQ8Bmc9ndIaL6YrKPy7TejqufvZrsCVzWxbQowpS3j2d03/Ujp2nXnH9rs2bOZOXMmH3/8MV26dGHs2LEh+zIkJibuWI6NjaWqqioisXZEW6trmbGomNfmF/LfZRuprVf2ze7KgJwQHxYtcNg+3Rmcm8bg3FQG9UghNante7THxcbs+OAP+XXPb6ow7Zew4j33AbvvMeEdl9AVRl7oHiVLXaJY8CwseRNScmHEBTDyIui+rz9xFy10dwVl38L4O+CQHzd/V5CcCZOehQ/vgll/dMef9xRkDfAnvo3LYdPy1p0jKR36HtY28QTZqyup9wapqalUVoaevbG8vJzMzEy6dOnC4sWL+eSTTyIcXeewvbaeD5aW8NqCdcxYVESgpp689CR+cFR/JgzPY1jPNOtHEo6P/gHz/g1H/sJ9+98T2YPghP+DY2+CpdNdsvjfPe7DuN9Rrghq2ASIb6M6tgXPwevXQnIGXPYm9Dk0vONiYuDo66HXQfCfH8AjY+GM+2HY6W0TV/UWWPQqfP4ErP209efrNRqueLf152nEEoTPunfvzhFHHMH+++9PcnIyPXrsLHsdP348Dz30EEOHDmXw4MEcemiYf7xmt+rrlU9XlTJ1QSHTviqivKqGjC7xnD0qn9NH9GJ038w2byHUoS16DWbcCPudCeNubP35YuNh6KnuUbEO5j8DXzwFr0x2dykHnuuSRd6IPTt/bTW8fQPMneISzzlTICWn5efZ9xj40QfwwqXwwiVw2NVw3K17ViymCgVzXb3Mwpdh+xZXF3L8bdD3yNbVdfhUBNZh5qTeXSV1Z9DZXm9jqsrX6yp4bX4hry9YT1FFgC4JsZwwrAcTRuRx5IDsFjeHNLgPtcdOgdwD4dKpbfftvrH6eljzIXz+JHwzFWoDkHsAjLzEJYzkMAdUKFvr6kkK58ER18C4m1pfz1G7Hab/Fub8E/oeAef8O/yK9q2b4Mvn3Osq+Qbiu7h6kFEXQ+9Dol4J3lwltSWIDqSzvV5wFbfLS7YwfWExry0oZGXJVuJjhe8NymbCiF4cNzSHLgl2o7zHNq+BR49131B/+C50zYrMdas2w1cvuSKYoi8hNtEVPY282N0RxDSR6Je/C//5oatMP/NBGHpa28b15Qsw9WeQlAbnPgZ9Dw+9X30drJzl4l88DeprXDHQqItdckgK3RAhGqLSismYtqSqrC8PsKS4kiVFlTv6FazYsIXtdfWIwCH9u3HFUftw0v65ZHRJiHbIe7+qMnhmItRthwumRS45gLtbGHOFe6yb7+oqvnwRvnoRMvu5Su0RF0Jantu/vh7++zdXqZwzFCY+6U+l8oETXS/x5y+Gx0519SmHXrnzLmDzGpj/NHzxNFQUuN7lY65wia3HsLaPx2eWIEz01dVCTOyOf7LgXsYNHcuWFFVSEajdcUhuWhKDc1M5emAWg3NTOWzf7vRMj1LnwdrtUN3KPjbJ3Zr+VhwNdTWuzH3TCrj4FVe5HC15I9zjhD/AN6+7b+Xv/QFm/ck1kx1+vquMXjYdDpgIp93jb7PUHvvB5Fmu2ez037pK5qETXB1Kw5hO+x7jkseQUyAusdnTtWeWIEx01QSovns4W+vieC/5BJ7YdjhfVnTZsTk1MY5BuamcOjzPdSzr4TqZtZs7hJWzXZHG1pLWnSct3zUDHXEhZPZtk9D2mCq88XNY9T6c8SD0Pyq68TSIT3bf4A+c6BJXwzf1Ze9ATDycfCcc/MPIlOknpbumrx/dCzNvcZX46b3he792v8eM9jk2XEtZHUQHsje+3vdefoRxX/6Sb7QPQ+Vb6ohhbfcjqBg6ie4jJ5DXLbV9NkGtr4f/3e2+yXYf2LoPpvpaWDbD9S8A2GesK6secmp0vn3+9y5491Y4+lcw7neRv35L1NXC6g8gNQ9yhkQnhqKFsG0T9DvS3QnvZawOwrRL//pwFX2+eJbNCd3p/6t5sGUtsV88Rb/5z8CHV8EXt8KI810rFr86Ke2JqjJ49SewZFrzvXJb4tCfuI5cDc09X/q+K4c/cFKbjo66WwtfdsnhgHPhmN9G5pqtERsH+46Lbgy53xlJqMNoR4WeHVPDaK574p577mHbtm1tHFH78Oh/V/LAGx8zLnYBaWMuJCkxwfWkPe5m+PnXcP5zkH8wfHQf3HcQTDnJfXhu3xrdwNd/CY98zxVrjP+za1/f2uTQIKMPjL0BrlkAF73s7iTm/gsePBweOca16Q+07Xhiu1j7mRtios9hMOG+qDe/NNFnCcJnliC+69H/ruQPb37Dr/MXEksdsSMajcoSGweDT4Lzn4FffAPH3QJbit239jsHw+vXQME8vjMsqd/mPwP/Ot51wrrsTTh0N0M27KmYWBhwrGtGed0SNzxEbcDVC9w5CF75Caz5qG1ff+kqeHYSpPeC856G+KS2O7fZa1kRk8+Ch/s+/vjjycnJ4YUXXqC6upozzzyTW2+9la1btzJx4kQKCgqoq6vjxhtvpLi4mHXr1nHMMceQlZXFrFmzov1S2sQjH6zgT9MWc8oBPTm38kPoObz55n+pPeDIn8MR18K3H7vORgueh3mPQc5+rvjlwPOgSzf/gq4JwNu/dtdsTa/cPdGlmyt+OuTHu46OuuCZthsdtWozPH0uaL0bgK9r990fYzqFzlNJ/dYNUPRV21409wA46Y5mdwke7vudd97hpZde4uGHH0ZVmTBhAr/61a8oKSnh7bff5p///CfgxmhKT0/fMeR3VlZ47c/beyX1w++v4Pa3FnPKgT35+zEJxD18hCumOfTHLTtRoBwW/scli3WfQ2yCa0446hLoP7Ztm4uWfeuae677wiWpcTdGZvTR5mzfunN01G8/BomFtF6wpzcz1VvcsA+XvNZ0xy/TYVkldTvxzjvv8M477zBy5EgAtmzZwrJlyzjqqKO47rrr+PWvf82pp57KUUe1k2aFbeih91dwx1uLOW14HndPHE7cuzdDTBzsf3bLT5aUDqO/7x5FC71OVM+7eZHT++xsLprRe/fnas7yma4Ja32dK3YZemrrztdWgkdH3bjMjYxasa515zzgHEsO5js6T4LYzTf9SFBVfvOb3/CjH/3oO9s+//xzpk2bxu9//3uOPfZYbrrppihE6I8HZi/nL28vYcLwPO6aOJw4UTdkwYDjISW7dSfP3R9O+rMbQG3xGy5ZzL4dZt/hWreMuhgGn9yy5qL19fDBX9w5coa5iWX8Goq6tbIGupFRjfFB50kQURI83PeJJ57IjTfeyIUXXkhKSgqFhYXEx8dTW1tLt27duOiii8jIyODRRx/d5dhwi5jao/tnLeev05dw+og8/nbucDfp+/J3YUuRa8LaVuKT3LfgA87ZdbiDFy9zvZSHTwpvuINtpfDyFe7u4cBJcOrdkNCl+WOM6aAsQfgseLjvk046iQsuuIDDDnMTe6SkpPDUU0+xfPlyfvnLXxITE0N8fDwPPvggAJMnT2b8+PHk5eXtlZXU9723jDvfWcoZI/L428QRxDYMr73gWVdMNGi8PxfO7Ova8H/v17BilqvY/eyf8MkDOwdM2/9sSEzd9bh1X8Dzl7jkdcpdrgjLmnqaTqzzVFJ3Au3p9f7j3WX8bcZSzhrZi7+eO3xncghUuKaaI853384jZetGN17PF09CyWJvyOUzXcV270Pg88fdPARdc2DiE26qTGM6AaukNhH195nLuHvmUs4a1Yu/nhOUHMAb578Khl8Q2aC6ZsHhV8NhV+06acv8pyG1J1Suh32OgbP/Zc08jfFYgjBt6p6ZS7ln5jLOHpXPX845cNfkADD/Wei2L+RHZVZlV2TU+2D3OPF21/Jp0auuOOmo6/bKsXSM8UuHTxCq2j4He2tj7aGo8O4ZS/n7u8s496B87jg7RHLYvMbNGHbM79tH2X5iiquPGHVxtCMxpl3q0ENtJCUlsWnTpnbx4eknVWXTpk0kJUVneARV5S4vOUwcnc+fQyUHcE1bwQ3XbIxp93y9gxCR8cDfgVjgUVW9o9H2vsAUIBsoBS5S1QJvWx3Q0PX5W1Wd0NLr5+fnU1BQQElJK8fqb8dUlaqaOjQmgaEDIj+PwNrSbfz7f6uZ8r9VnDe6N7efdQAxoZKDqmu91O+o6M93YIwJi28JQkRigfuB44ECYI6ITFXVRUG73Qk8oaqPi8g44Hag4X6/SlVHtCaG+Ph4+vfv35pTtFs1dfW8OLeAe99dRlFFAIAuCas4YkAW44bkcMzgHHLT2/6OoqaunrmrNzNryQbeW7yB5Ru2AHDBIX34w+n7h04OAAVzoHQFHPWLNo/JGOMPP+8gxgDLVXUlgIg8B5wOBCeIYUDDJ8Ys4FUf4+kQ6uuVN75az90zlrJq41ZG9sngznOHs72ujvcWb2DW4hJmLCoGYGjPNMYNyWbckBxG9M4MXewThpLKamYv2cCsJRv479KNVFbXkhAbwyH7dOOCMX0YNySHflm7meJxwbMQl+ymZjTG7BX8TBC9gLVBzwuAQxrtswA4C1cMdSaQKiLdVXUTkCQic4Fa4A5VfbXxBURkMjAZoE+fjjHFX1NUldlLSvjr9CUsWl/BkNxUHr1kNMcOzdlRCT9uSA9UlaXFW7xksYGH3l/J/bNWkNklnu8NyuaYITl8b1B2s1N21tcrXxWWM2uJO8eCgnIAeqQlcsqBPTlmSA5HDsiia2KYfz41ATe43tDTICmt1e+FMSYyot2K6XrgPhG5DPgAKATqvG19VbVQRPYB3hORr1R1RfDBqvoI8Ai4jnKRCzuyPltVyl+nL2bO6s306daFe84bwYTheSGLc0SEwblu3uafjN2X8m01fLCshFmLNzB7aQmvzl9HjMCoPpkcMySHcUNyGJKbSmV1LR8u28h7izcwe0kJG7dUIwIjemdw3fGDOGZIDvvlpe1Zi7Clb7sRWIdPaoN3wxgTKX4miEIgeDjNfG/dDqq6DncHgYikAGerapm3rdD7uVJEZgMjgV0SREe3sLCcO99ZwuwlJeSkJvKHM/bnvIN7Ex8bfuOz9C7xnDY8j9OG51FXrywoKGPWYld/8NfpS/jr9CVkpyayeet2auuVtKQ4vjc4h3FDsjl6YDbdU9pgTuQFz7nOaPuMbf25jDER42eCmAMMFJH+uMQwCdil+6yIZAGlqloP/AbXogkRyQS2qWq1t88RwF98jLVdWVmyhbtmLOWNL9eTnhzPDScN4dLD+pGc0LpOXLExwqg+mYzqk8l1JwymuCLA7CUb+N/yTeRlJDNuSA6j+mS4AfXaypYSWD7D9WC2TmjG7FV8SxCqWisiVwPTcc1cp6jq1yJyGzBXVacCY4HbRURxRUxXeYcPBR4WkXpcX407GrV+6pDWlVVx77vLeHFeAYlxMfx03ACuOHof0pLifblej7Qkzju4D+cd7GP9zcKXoL7WjYxqjNmrdOjB+vYWm7ZU88DsFTz5yRpQuPDQPlw5dgDZqW1QvBNtDx/tfv7og+jGYYwJyQbra6cqAzU8+t9V/OvDVWzbXsvZo/K55riB5Gd2kPkHihfB+gVuWlFjzF7HEkQUBGrqeOqTNdw/azmbt9Vw8gG5/OL4QQzISd39wXuTBc/u+bSixpioswQRQbV19bw4r4C/z3S9n48amMWvThzCAfnp0Q6t7dXXtd20osaYqLAEEQH19cqbX63nrqDez3efN4LD9u3A8w6snNX204oaYyLKEoSPwun93GEteM7faUWNMb6zBOGTOatL+cvbu/Z+Pm143h6Ph7RXCVTAN2+4u4e4DtASy5hOyhJEG2uL3s97vUWvedOKWvGSMXszSxBtxK/ez3ulBc9504oeHO1IjDGtYAmilVSVP775Df/+aPWO3s8/PGof0pP96f3c7rW3aUWNMXvMEkQrFVdU8+iHqzj5gFxunbB/x+j93Bo2ragxHUYnKhj3x/ryKgDOHpVvycGmFTWmQ7EE0UrF3nSfPdLafnrPvU7DtKI274MxHYIliFYqKncJwo/5n8OyfRu8cCnM+Zf7Bh9NNq2oMR2KJYhWKqqoJj5W6NbMFJ6+mn07LHoV3vwFvPIj2L41OnHYtKLGdDhWSd1KxRUBclKTQk7/6bt1X8DH98GIi6BbP3jvj1D0FUx8ErIGRDYWm1bUmA7H7iBaqag8EJ3ipboaeO2n0DUbTvwDHP1LuPhlqCyCR8bCoqmRjcemFTWmw7EE0UrFFQFyo1FB/dG9UPwVnHwnJGe6dfuOcxPzZA+CFy6Gd26Eulr/Y2mYVvTAiTatqDEdiCWIVlBViioCkW/BtHEZzP6zK+8f1qhCOKM3XP4WjP6BSyJPnA6Vxf7GY9OKGtMhWYJohcrqWrZtryM3PYL9H+rrYerPID7J3T2EEpcIp94FZz4MhfPctJ/ffuJPPMVfw9wp0HM49BjmzzWMMVFhCaIVisuj0Adi3r/h24/ghD9Cam7z+w6fBD+cCQld4LFT4OMH2qYpbKAC5v4bHjkGHjwcSlfBkT9v/XmNMe2KtWJqhSKvk1zE6iDKC2HGzdD/ezDyovCOyd0fJs+GV6+E6b+Bgs9gwj8gsYXTm6rCtx/D50+6ZrU12yBnGJx4Oxx4HnTtwJMfGdNJWYJohYh2klOFN69zZf2n/b1lA+ElpcN5T8H//g7v3grFi+C8JyF78O6PrSx2HeC+eBI2LYeEFDjgXBh1CfQ6yAbkM6YD87WISUTGi8gSEVkuIjeE2N5XRN4VkS9FZLaI5Adtu1RElnmPS/2Mc09FdJiNr1+GpW/BuN9Bt/4tP14EjrwWLnkNqkrhn+Ng4cuh962rhSVvw3MXwl1DYebN0CULTn8Arl8KE+6F/NGWHIzp4Hy7gxCRWOB+4HigAJgjIlNVdVHQbncCT6jq4yIyDrgduFhEugE3A6MBBeZ5x272K949UVQRIKNLPEnxPjft3FYK034FeSPhkJ+07lz9j3ZNYV+4FF663I2fdPxtEBsPm1bAF0+5O4bK9a6PxWFXwciLXdNZY0yn4mcR0xhguaquBBCR54DTgeAEMQz4hbc8C3jVWz4RmKGqpd6xM4DxwLM+xttiReXVkal/mP5bCJTBhNcgtg1+ZWl5cNmbMONG+OQBKJjrWj6t/i9IDAw4Hk7+q5tPOraTzmthjPE1QfQC1gY9LwAOabTPAuAs4O/AmUCqiHRv4thejS8gIpOByQB9+vRps8DDVRyJPhDLZ7pv9Edd7yqc20pcApz0Zzfr2+vXukrmcTfCiAtcAjHGdHrRrqS+HrhPRC4DPgAKgbpwD1bVR4BHAEaPHh3xoUyLKgIM6+njwHTVW+D1n0P3gW4oDT8ccA7sdyYgEGOtno0xO/mZIAqB3kHP8711O6jqOtwdBCKSApytqmUiUgiMbXTsbB9jbbGauno2bqmmh58tmN77A5R/C5e/7TrG+cWGxzDGhODnV8Y5wEAR6S8iCcAkYJcR5EQkS0QaYvgNMMVbng6cICKZIpIJnOCtazdKKqtR9bEPxNo58OlDcPAPoe9h/lzDGGOa4VuCUNVa4GrcB/s3wAuq+rWI3CYiDQMIjQWWiMhSoAfwR+/YUuD/cElmDnBbQ4V1e7Gjk5wfw2zUboepP3V1Acfe3PbnN8aYMPhaB6Gq04BpjdbdFLT8EvBSE8dOYecdRbuzyzAbWze5SXLaqsXPh3dByTdwwQs2+Y4xJmqiXUm912q4g+iZWAP3HAgJXWHE+a7PQNbAPT/xhm/ggztdb+VBJ7ZRtMYY03LWbGUPFVUESIiNITOwFmq2QkoP+Og+uG80TBkP859p+fSf9XWuaCkxFcbf4U/gxhgTJksQe6i4PEBOWiKyeZVbceZD8Itv4LhbYWsJvPoTuHMwvH4NFMwLbxTVzx5xPZtP+jN0zfL3BRhjzG5YEdMeKmqYSW7zarcisx8kprjxjo64ZufIpwueh3mPQc5+MOpiN/Jpl27fPeHmNfDuba4X8wHnRu6FGGNME+wOYg8VV3h9IDavgq45Ljk0EIG+h8OZD8L1S+DUu91QFm/fAH8bDC9eDivec5P/gLu7eONaN8zFqXfbIHjGmHbB7iD2gKpSVB5g3JAc2LjK3T00JSkdRn/fPYq/dncVXz7nRmdN7wMjL3QV3Cveg5P+6qYMNcaYdsASxB6oCNRSVVPnipiWrXZ3C+HosR+cdAccfyssfsMli9l3AAq9D3Gd4owxpp2wBLEHGuaByE2NgfICyGzh/AxxibD/2e6xeQ188zoMO93GQjLGtCuWIPZAw0xyfWUjoHs2gU+DzL5w+NVtE5gxxrQh+8q6B3YMs1G33q1org7CGGP2UpYg9kDDMBuZ1d7gtC0tYjLGmL1AWAlCRF4WkVOCRl7t1IoqAmR2iSe+4luI7wopOdEOyRhj2ly4H/gPABcAy0TkDhEZ7GNM7d6OmeQ2e01crd+CMaYDCitBqOpMVb0QGAWsBmaKyEcicrmIdLpJi4sqAuSmJ0HpbvpAGGPMXizsIiNvrujLgB8CX+DmkR4FzPAlsnasqLya3NREN8xGa1owGWNMOxZWM1cReQUYDDwJnKaqXvMdnheRuX4F1x7V1NWzaWs1+yRvhdoqu4MwxnRY4faDuFdVZ4XaoKqj2zCedm+DN9Vo/9gNboXdQRhjOqhwi5iGiUhGwxNvrugr/QmpfWvoJJevRW6FNXE1xnRQ4SaIK1S1rOGJqm4GrvAlonauYZiNrJp1bvTVdBtczxjTMYWbIGJFdrblFJFYIMGfkNq3hjuItKpCSM+HuE75NhhjOoFw6yDexlVIP+w9/5G3rtMprgiQEBdDQuUaK14yxnRo4SaIX+OSwk+85zOAR32JqJ0rqgjQIy0RKV0FQ06JdjjGGOObcDvK1avqg6p6jvd4WFXrdneciIwXkSUislxEbgixvY+IzBKRL0TkSxE52VvfT0SqRGS+93io5S/NH+vLA/RLqYdtG60FkzGmQwu3H8RA4HZgGJDUsF5V92nmmFjgfuB4oACYIyJTVXVR0G6/B15Q1QdFZBgwDejnbVuhqiPCfymRUVwRYHxWqXtiRUzGmA4s3ErqfwMPArXAMcATwFO7OWYMsFxVV6rqduA54PRG+yiQ5i2nA+vCjCcqGqYaHRC30a2wOwhjTAcWboJIVtV3AVHVNap6C7C7AvhewNqg5wXeumC3ABeJSAHu7uGnQdv6e0VP74vIUaEuICKTRWSuiMwtKSkJ86XsufKqGqpr6+kjxW6F9aI2xnRg4SaIam+o72UicrWInAmktMH1zwceU9V84GTgSe8664E+qjoS+AXwjIikNT5YVR9R1dGqOjo7O7sNwmneLhMFJXeDpHTfr2mMMdESboK4BugC/Aw4CLgIuHQ3xxQCwb3I8r11wX4AvACgqh/j6jeyVLVaVTd56+cBK4BBYcbqm4Y+EBnVhVa8ZIzp8HabILzK5vNUdYuqFqjq5ap6tqp+sptD5wADRaS/iCQAk4Cpjfb5FjjWu85QXIIoEZFs77qIyD7AQGBli16ZDxp6UXfdstYqqI0xHd5uE4TXnPXIlp5YVWuBq4HpwDe41kpfi8htIjLB2+064AoRWQA8C1ymqgocDXwpIvOBl4Afq2ppS2Noa0Xl1cRRS2xlgdU/GGM6vHA7yn0hIlOBF4GtDStV9eXmDlLVabjK5+B1NwUtLwKOCHHcf4D/hBlbxBRVBNivSwVSX2dFTMaYDi/cBJEEbALGBa1ToNkE0dEUVwTYv8tm2IIVMRljOrywEoSqXu53IHuDovIAYxOsD4QxpnMItyf1v3F3DLtQ1e+3eUTtWHFFgH6ZGyA2EVJyox2OMcb4KtwipjeClpOAM2nnvZ7bWnVtHZu2bicvo8hVUMeEPZ23McbslcItYtqlwlhEngU+9CWidmpDRTUA3WvWQa4VLxljOr49/Ro8EMhpy0DaO9cHQkndVmAV1MaYTiHcOohKdq2DKMLNEdFpFFUE6E4FcbVbrQ+EMaZTCLeIKdXvQNq7ovIAfRsG6bMWTMaYTiCsIiYROVNE0oOeZ4jIGb5F1Q4VVwTYp2GYbytiMsZ0AuHWQdysquUNT1S1DLjZl4jaqaKKaoYlbQIEMvpEOxxjjPFduM1cQyWScI/tEIrLA+wbVwLJeRCftPsDjDFmLxfuHcRcEblLRPb1HncB8/wMrL0pqgiQT5EVLxljOo1wE8RPge3A87ipQwPAVX4F1d6oKkUVAXJqi6Bbv2iHY4wxERFuK6atwA0+x9JulW2rIaa2itSajXYHYYzpNMJtxTRDRDKCnmeKyHTfompniioC9JEN7on1gTDGdBLhFjFleS2XAFDVzXSintRFFdYHwhjT+YSbIOpFZEfbThHpR4jRXTuq4vLgOwhLEMaYziHcpqq/Az4UkfcBAY4CJvsWVTvjipiK0aR0pEu3aIdjjDEREdYdhKq+DYwGluDmjr4OqPIxrnaluCLAgPgSxOofjDGdSLiD9f0QuAbIB+YDhwIfs+sUpB1WUXmAfrIBMsdEOxRjjImYcOsgrgEOBtao6jHASKDMr6Damw3l2+hRv8EqqI0xnUq4CSKgqgEAEUlU1cXAYP/Camcq1hFHrVVQG2M6lXATRIHXD+JVYIaIvAas2d1BIjJeRJaIyHIR+U5HOxHpIyKzROQLEflSRE4O2vYb77glInJimHG2ueraOtICBe6J1UEYYzqRcHtSn+kt3iIis4B04O3mjhGRWOB+4HigAJgjIlNVdVHQbr8HXlDVB0VkGDAN6OctTwL2A/KAmSIySFXrWvDa2sSGimrrA2GM6ZRaPOWoqr6vqlNVdftudh0DLFfVld6+zwGnNz4dkOYtpwPrvOXTgedUtVpVVwHLvfNFXEMv6vqYeEjrFY0QjDEmKvZ0Tupw9ALWBj0v8NYFuwW4SEQKcHcPP23BsYjIZBGZKyJzS0pK2iruXRSVuz4Qtam9ISbWl2sYY0x75GeCCMf5wGOqmg+cDDwpImHHpKqPqOpoVR2dnZ3tS4DF3jAbYsVLxphOxs8EUQj0Dnqe760L9gPgBQBV/RhIArLCPDYiisqq6CsbiMvaJxqXN8aYqPEzQcwBBopIfxFJwFU6T220z7fAsQAiMhSXIEq8/SaJSKKI9AcGAp/5GGuTKspKSJNtdgdhjOl0fJs2VFVrReRqYDoQC0xR1a9F5DZgrqpOxQ3Z8U8R+TmuwvoyVVXgaxF5AVgE1AJXRaMFE0BM2Sq3YH0gjDGdjK/zSqvqNFzlc/C6m4KWFwFHNHHsH4E/+hlfOJIqv3UL1gfCGNPJRLuSul1TVdKqvKoPSxDGmE7GEkQzNm+roZcWsy0hGxK6RDscY4yJKEsQzSgqD9A3pphAap/d72yMMR2MJYhmFDdMFJTRN9qhGGNMxFmCaMaGzeXkspn47H2jHYoxxkScJYhmVG1YRYwoXXIHRDsUY4yJOEsQzdDSlQDEdbc7CGNM52MJohnxFd6UF9bE1RjTCVmCaEbK1rUEJBm6ZkU7FGOMiThLEM3I3L6OzYm9QCTaoRhjTMRZgmhCoKaOvPoitnbtvfudjTGmA7IE0YTi8m30kQ3UplsfCGNM52QJogml69eQKDXEdrd5IIwxnZMliCZsLV4BQHIP6wNhjOmcLEE0oXajSxDpeQOjHIkxxkSHJYgmxJatoVZjSMnpF+1QjDEmKixBNCF5y7cUx+QgcQnRDsUYY6LCEkQT0gKFbErIi3YYxhgTNZYgmpBTu44tyfnRDsMYY6LGEkQI9dvKyKDSJgoyxnRqliBCKF+/zC1k9o9uIMYYE0WWIEKoXOcSRGKODfNtjOm8fE0QIjJeRJaIyHIRuSHE9rtFZL73WCoiZUHb6oK2TfUzzsa2l7g+ECk9rZOcMabzivPrxCISC9wPHA8UAHNEZKqqLmrYR1V/HrT/T4GRQaeoUtURfsXXHN28io2aRk6WDfNtjOm8/LyDGAMsV9WVqrodeA44vZn9zwee9TGesCVWrOFbzSE7JTHaoRhjTNT4mSB6AWuDnhd4675DRPoC/YH3glYnichcEflERM7wLcoQUrYVUBybR1ysVdEYYzov34qYWmgS8JKq1gWt66uqhSKyD/CeiHylqiuCDxKRycBkgD592qhJau12Mmo2UJ48tm3OZ4wxeyk/vyIXAsGz7eR760KZRKPiJVUt9H6uBGaza/1Ewz6PqOpoVR2dnZ3dFjFD+VpiqGdbivWBMMZ0bn4miDnAQBHpLyIJuCTwndZIIjIEyAQ+DlqXKSKJ3nIWcASwqPGxvihdBUBder+IXM4YY9or34qYVLVWRK4GpgOxwBRV/VpEbgPmqmpDspgEPKeqGnT4UOBhEanHJbE7gls/+alm40rigbgsmyjIGNO5+VoHoarTgGmN1t3U6PktIY77CDjAz9iasq14OQmaQGqWjcNkjOncrJlOI3WbVvKt5pCbnhztUIwxJqosQTQSV76ab7UHuenWB8IY07lZggimSvKWtazRHHqkJUU7GmOMiSpLEMG2bCC+PkBxTC6pSfHRjsYYY6LKEkSwza6Ja2XX3rvZ0RhjOj5LEMG8PhDbU/tGORBjjIk+SxDBNq+mHiG2myUIY4yxBBFES1eyXruTlZEa7VCMMSbqLEEEqd24kjX1OeRaCyZjjLEEEUzKVrNGe1gTV2OMwRLETtVbiKva6HWSswRhjDGWIBpsXg3ghtmwOwhjjLEEsYPXB2ItOWSlJEQ5GGOMiT5LEA28PhBbu/axqUaNMQZLEDttXk1lTCop6d2jHYkxxrQLliAabF7FOrEWTMYY08ASRIPSVaysy7EWTMYY47EEAVBXi5avZUVttt1BGGOMxxIEQPlapL6WNdrDmrgaY4zHEgTs6AOxVq2IyRhjGliCgB19INbUWyW1McY0sAQBULqK2pgEisi0OwhjjPH4miBEZLyILBGR5SJyQ4jtd4vIfO+xVETKgrZdKiLLvMelfsbJ5lVsju9J18QEUhLjfL2UMcbsLXz7NBSRWOB+4HigAJgjIlNVdVHDPqr686D9fwqM9Ja7ATcDowEF5nnHbvYl2M2rKYrNpUdaoi+nN8aYvZGfdxBjgOWqulJVtwPPAac3s//5wLPe8onADFUt9ZLCDGC8L1GqQqkb5tuKl4wxZic/E0QvYG3Q8wJv3XeISF+gP/BeS49ttW2bYHsly7ZnWQW1McYEaS+V1JOAl1S1riUHichkEZkrInNLSkr27MrxXaib+CRTqw60PhDGGBPEzwRRCPQOep7vrQtlEjuLl8I+VlUfUdXRqjo6Ozt7z6JM6MKm/BNYVW99IIwxJpifCWIOMFBE+otIAi4JTG28k4gMATKBj4NWTwdOEJFMEckETvDW+aKoIgBgRUzGGBPEt1ZMqlorIlfjPthjgSmq+rWI3AbMVdWGZDEJeE5VNejYUhH5P1ySAbhNVUv9irWo3CUIK2IyxpidfG30r6rTgGmN1t3U6PktTRw7BZjiW3BBir07CCtiMsaYndpLJXVUFVUEiI0RslKsH4QxxjSwBAEUlVeTnZJIbIxEOxRjjGk3LEHgiph6WPGSMcbswhIErogp14bZMMaYXViCAIrLA9aCyRhjGun0CWJrdS2V1bVWxGSMMY10+gRRXVvPacPz2D8vPdqhGGNMu9LpJz/o1jWBf5w/MtphGGNMu9Pp7yCMMcaEZgnCGGNMSJYgjDHGhGQJwhhjTEiWIIwxxoRkCcIYY0xIliCMMcaEZAnCGGNMSBI0kdteTURKgDWtOEUWsLGNwvGDxdc6Fl/rWHyt057j66uq2aE2dJgE0VoiMldVR0c7jqZYfK1j8bWOxdc67T2+plgRkzHGmJAsQRhjjAnJEsROj0Q7gN2w+FrH4msdi6912nt8IVkdhDHGmJDsDsIYY0xIliCMMcaE1KkShIiMF5ElIrJcRG4IsT1RRJ73tn8qIv0iGFtvEZklIotE5GsRuSbEPmNFpFxE5nuPmyIVX1AMq0XkK+/6c0NsFxG513sPvxSRURGMbXDQezNfRCpE5NpG+0T0PRSRKSKyQUQWBq3rJiIzRGSZ9zOziWMv9fZZJiKXRjC+v4rIYu/394qIZDRxbLN/Cz7Gd4uIFAb9Dk9u4thm/999jO/5oNhWi8j8Jo71/f1rNVXtFA8gFlgB7AMkAAuAYY32uRJ4yFueBDwfwfh6AqO85VRgaYj4xgJvRPl9XA1kNbP9ZOAtQIBDgU+j+PsuwnUCitp7CBwNjAIWBq37C3CDt3wD8OcQx3UDVno/M73lzAjFdwIQ5y3/OVR84fwt+BjfLcD1Yfz+m/1/9yu+Rtv/BtwUrfevtY/OdAcxBliuqitVdTvwHHB6o31OBx73ll8CjhURiURwqrpeVT/3liuBb4Bekbh2GzsdeEKdT4AMEekZhTiOBVaoamt617eaqn4AlDZaHfx39jhwRohDTwRmqGqpqm4GZgDjIxGfqr6jqrXe00+A/La+briaeP/CEc7/e6s1F5/32TEReLatrxspnSlB9ALWBj0v4LsfwDv28f5ByoHuEYkuiFe0NRL4NMTmw0RkgYi8JSL7RTYyABR4R0TmicjkENvDeZ8jYRJN/2NG+z3soarrveUioEeIfdrL+/h93B1hKLv7W/DT1V4R2JQmiujaw/t3FFCsqsua2B7N9y8snSlB7BVEJAX4D3CtqlY02vw5rshkOPAP4NUIhwdwpKqOAk4CrhKRo6MQQ7NEJAGYALwYYnN7eA93UFfW0C7bmovI74Ba4OkmdonW38KDwL7ACGA9rhinPTqf5u8e2v3/UmdKEIVA76Dn+d66kPuISByQDmyKSHTumvG45PC0qr7ceLuqVqjqFm95GhAvIlmRis+7bqH3cwPwCu5WPlg477PfTgI+V9Xixhvaw3sIFDcUu3k/N4TYJ6rvo4hcBpwKXOglse8I42/BF6parKp1qloP/LOJ60b7/YsDzgKeb2qfaL1/LdGZEsQcYKCI9Pe+YU4CpjbaZyrQ0FrkHOC9pv452ppXXvkv4BtVvauJfXIb6kREZAzu9xfJBNZVRFIblnGVmQsb7TYVuMRrzXQoUB5UnBIpTX5zi/Z76An+O7sUeC3EPtOBE0Qk0ytCOcFb5zsRGQ/8Cpigqtua2CecvwW/4guu0zqzieuG8//up+OAxapaEGpjNN+/Fol2LXkkH7gWNktxrRt+5627DfePAJCEK5ZYDnwG7BPB2I7EFTV8Ccz3HicDPwZ+7O1zNfA1rkXGJ8DhEX7/9vGuvcCLo+E9DI5RgPu99/grYHSEY+yK+8BPD1oXtfcQl6jWAzW4cvAf4Oq13gWWATOBbt6+o4FHg479vve3uBy4PILxLceV3zf8HTa07MsDpjX3txCh+J70/ra+xH3o92wcn/f8O//vkYjPW/9Yw99c0L4Rf/9a+7ChNowxxoTUmYqYjDHGtIAlCGOMMSFZgjDGGBOSJQhjjDEhWYIwxhgTkiUIY9oBb5TZN6IdhzHBLEEYY4wJyRKEMS0gIheJyGfeGP4Pi0isiGwRkbvFzePxrohke/uOEJFPguZVyPTWDxCRmd6AgZ+LyL7e6VNE5CVvLoanIzWSsDFNsQRhTJhEZChwHnCEqo4A6oALcb2356rqfsD7wM3eIU8Av1bVA3E9fxvWPw3cr27AwMNxPXHBjeB7LTAM19P2CJ9fkjHNiot2AMbsRY4FDgLmeF/uk3ED7dWzc1C2p4CXRSQdyFDV9731jwMveuPv9FLVVwBUNQDgne8z9cbu8WYh6wd86PurMqYJliCMCZ8Aj6vqb3ZZKXJjo/32dPya6qDlOuz/00SZFTEZE753gXNEJAd2zC3dF/d/dI63zwXAh6paDmwWkaO89RcD76ubLbBARM7wzpEoIl0i+SKMCZd9QzEmTKq6SER+j5sFLAY3gudVwFZgjLdtA66eAtxQ3g95CWAlcLm3/mLgYRG5zTvHuRF8GcaEzUZzNaaVRGSLqqZEOw5j2poVMRljjAnJ7iCMMcaEZHcQxhhjQrIEYYwxJiRLEMYYY0KyBGGMMSYkSxDGGGNC+n9SotiO9lzoiwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_loss_accuracy(history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "_-HIT6riyMfN"
   },
   "source": [
    "#### Model 9: Mobilenet (re train all weights) + GRU (128 cells) + 128 dense nodes + Random data transformations on training data set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "50W_MDsv4_JU"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# epochs = 15\n",
      "# batch size = 10\n",
      "# num frames per video = 30\n",
      "WARNING:tensorflow:`input_shape` is undefined or non-square, or `rows` is not in [128, 160, 192, 224]. Weights for input shape (224, 224) will be loaded as the default.\n",
      "Number of layers in the pretrained mobilenet model:  86\n",
      "Model: \"sequential_7\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " time_distributed_4 (TimeDis  (None, 30, 3, 3, 1024)   3228864   \n",
      " tributed)                                                       \n",
      "                                                                 \n",
      " time_distributed_5 (TimeDis  (None, 30, 3, 3, 1024)   4096      \n",
      " tributed)                                                       \n",
      "                                                                 \n",
      " time_distributed_6 (TimeDis  (None, 30, 1, 1, 1024)   0         \n",
      " tributed)                                                       \n",
      "                                                                 \n",
      " time_distributed_7 (TimeDis  (None, 30, 1024)         0         \n",
      " tributed)                                                       \n",
      "                                                                 \n",
      " gru_1 (GRU)                 (None, 128)               443136    \n",
      "                                                                 \n",
      " dropout_14 (Dropout)        (None, 128)               0         \n",
      "                                                                 \n",
      " dense_20 (Dense)            (None, 128)               16512     \n",
      "                                                                 \n",
      " dropout_15 (Dropout)        (None, 128)               0         \n",
      "                                                                 \n",
      " dense_21 (Dense)            (None, 5)                 645       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 3,693,253\n",
      "Trainable params: 3,669,317\n",
      "Non-trainable params: 23,936\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "num_epochs = 15 # choose the number of epochs\n",
    "print ('# epochs =', num_epochs)\n",
    "batch_size = 10  #experiment with the batch size\n",
    "print ('# batch size =', batch_size)\n",
    "num_frames = 30\n",
    "print ('# num frames per video =', num_frames)\n",
    "height = 120\n",
    "width = 120\n",
    "\n",
    "model = mobilenet_RNN(gru=True, num_images=num_frames)\n",
    "print(model.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "8WLYJzPJyLuB"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:`period` argument is deprecated. Please use `save_freq` to specify the frequency in number of batches seen.\n"
     ]
    }
   ],
   "source": [
    "callbacks_list = model_callbacks(\"model_retrain_mobilenet_gru\")\n",
    "\n",
    "steps_per_epoch, validation_steps = calculate_steps(num_train_sequences, num_val_sequences, batch_size)\n",
    "\n",
    "train_generator = generate(train_path, train_doc, batch_size, augment=True, num_images=num_frames)\n",
    "\n",
    "val_generator   = generate(val_path, val_doc, batch_size, num_images=num_frames)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "SpD988i9yLuE"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.9975 - categorical_accuracy: 0.6154\n",
      "Epoch 00001: val_loss improved from inf to 0.75179, saving model to model_retrain_mobilenet_gru_2023-08-2907_28_23.340432/model-00001-0.99746-0.61538-0.75179-0.73000.h5\n",
      "67/67 [==============================] - 198s 3s/step - loss: 0.9975 - categorical_accuracy: 0.6154 - val_loss: 0.7518 - val_categorical_accuracy: 0.7300 - lr: 0.0010\n",
      "Epoch 2/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3599 - categorical_accuracy: 0.8824\n",
      "Epoch 00002: val_loss improved from 0.75179 to 0.40959, saving model to model_retrain_mobilenet_gru_2023-08-2907_28_23.340432/model-00002-0.35987-0.88235-0.40959-0.83000.h5\n",
      "67/67 [==============================] - 188s 3s/step - loss: 0.3599 - categorical_accuracy: 0.8824 - val_loss: 0.4096 - val_categorical_accuracy: 0.8300 - lr: 0.0010\n",
      "Epoch 3/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3315 - categorical_accuracy: 0.9005\n",
      "Epoch 00003: val_loss did not improve from 0.40959\n",
      "67/67 [==============================] - 190s 3s/step - loss: 0.3315 - categorical_accuracy: 0.9005 - val_loss: 0.5426 - val_categorical_accuracy: 0.8200 - lr: 0.0010\n",
      "Epoch 4/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3161 - categorical_accuracy: 0.9020\n",
      "Epoch 00004: val_loss did not improve from 0.40959\n",
      "67/67 [==============================] - 184s 3s/step - loss: 0.3161 - categorical_accuracy: 0.9020 - val_loss: 0.8781 - val_categorical_accuracy: 0.7400 - lr: 0.0010\n",
      "Epoch 5/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2167 - categorical_accuracy: 0.9427\n",
      "Epoch 00005: val_loss improved from 0.40959 to 0.21287, saving model to model_retrain_mobilenet_gru_2023-08-2907_28_23.340432/model-00005-0.21668-0.94268-0.21287-0.91000.h5\n",
      "67/67 [==============================] - 188s 3s/step - loss: 0.2167 - categorical_accuracy: 0.9427 - val_loss: 0.2129 - val_categorical_accuracy: 0.9100 - lr: 0.0010\n",
      "Epoch 6/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1647 - categorical_accuracy: 0.9502\n",
      "Epoch 00006: val_loss improved from 0.21287 to 0.05735, saving model to model_retrain_mobilenet_gru_2023-08-2907_28_23.340432/model-00006-0.16470-0.95023-0.05735-0.99000.h5\n",
      "67/67 [==============================] - 187s 3s/step - loss: 0.1647 - categorical_accuracy: 0.9502 - val_loss: 0.0573 - val_categorical_accuracy: 0.9900 - lr: 0.0010\n",
      "Epoch 7/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1132 - categorical_accuracy: 0.9623\n",
      "Epoch 00007: val_loss did not improve from 0.05735\n",
      "67/67 [==============================] - 189s 3s/step - loss: 0.1132 - categorical_accuracy: 0.9623 - val_loss: 0.1777 - val_categorical_accuracy: 0.9400 - lr: 0.0010\n",
      "Epoch 8/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1086 - categorical_accuracy: 0.9668\n",
      "Epoch 00008: val_loss did not improve from 0.05735\n",
      "67/67 [==============================] - 188s 3s/step - loss: 0.1086 - categorical_accuracy: 0.9668 - val_loss: 0.2406 - val_categorical_accuracy: 0.9300 - lr: 0.0010\n",
      "Epoch 9/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1891 - categorical_accuracy: 0.9472\n",
      "Epoch 00009: val_loss did not improve from 0.05735\n",
      "\n",
      "Epoch 00009: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.\n",
      "67/67 [==============================] - 202s 3s/step - loss: 0.1891 - categorical_accuracy: 0.9472 - val_loss: 0.1004 - val_categorical_accuracy: 0.9800 - lr: 0.0010\n",
      "Epoch 10/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0629 - categorical_accuracy: 0.9864\n",
      "Epoch 00010: val_loss did not improve from 0.05735\n",
      "67/67 [==============================] - 208s 3s/step - loss: 0.0629 - categorical_accuracy: 0.9864 - val_loss: 0.0777 - val_categorical_accuracy: 0.9900 - lr: 1.0000e-04\n",
      "Epoch 11/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0649 - categorical_accuracy: 0.9804\n",
      "Epoch 00011: val_loss did not improve from 0.05735\n",
      "67/67 [==============================] - 205s 3s/step - loss: 0.0649 - categorical_accuracy: 0.9804 - val_loss: 0.0597 - val_categorical_accuracy: 0.9900 - lr: 1.0000e-04\n",
      "Epoch 12/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0550 - categorical_accuracy: 0.9834\n",
      "Epoch 00012: val_loss did not improve from 0.05735\n",
      "\n",
      "Epoch 00012: ReduceLROnPlateau reducing learning rate to 0.0001.\n",
      "67/67 [==============================] - 208s 3s/step - loss: 0.0550 - categorical_accuracy: 0.9834 - val_loss: 0.0639 - val_categorical_accuracy: 0.9800 - lr: 1.0000e-04\n",
      "Epoch 13/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0316 - categorical_accuracy: 0.9955\n",
      "Epoch 00013: val_loss did not improve from 0.05735\n",
      "67/67 [==============================] - 198s 3s/step - loss: 0.0316 - categorical_accuracy: 0.9955 - val_loss: 0.0947 - val_categorical_accuracy: 0.9800 - lr: 1.0000e-04\n",
      "Epoch 14/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0390 - categorical_accuracy: 0.9894\n",
      "Epoch 00014: val_loss improved from 0.05735 to 0.05449, saving model to model_retrain_mobilenet_gru_2023-08-2907_28_23.340432/model-00014-0.03903-0.98944-0.05449-0.99000.h5\n",
      "67/67 [==============================] - 187s 3s/step - loss: 0.0390 - categorical_accuracy: 0.9894 - val_loss: 0.0545 - val_categorical_accuracy: 0.9900 - lr: 1.0000e-04\n",
      "Epoch 15/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0226 - categorical_accuracy: 0.9925\n",
      "Epoch 00015: val_loss improved from 0.05449 to 0.01405, saving model to model_retrain_mobilenet_gru_2023-08-2907_28_23.340432/model-00015-0.02262-0.99246-0.01405-1.00000.h5\n",
      "67/67 [==============================] - 187s 3s/step - loss: 0.0226 - categorical_accuracy: 0.9925 - val_loss: 0.0140 - val_categorical_accuracy: 1.0000 - lr: 1.0000e-04\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(train_generator, steps_per_epoch=steps_per_epoch, epochs=num_epochs, verbose=1, \n",
    "                              callbacks=callbacks_list, validation_data=val_generator, \n",
    "                              validation_steps=validation_steps, class_weight=None, workers=1, initial_epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "RFNq8m6MyLuH"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['loss', 'categorical_accuracy', 'val_loss', 'val_categorical_accuracy', 'lr'])\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_loss_accuracy(history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Kv330voPpGim"
   },
   "source": [
    "#### Model 10: Mobilenet (re train all wieghts) + LSTM (128 cells) + 128 dense nodes\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "rv7aoMwOpCSb"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# epochs = 15\n",
      "# batch size = 10\n",
      "# num frames per video = 20\n",
      "WARNING:tensorflow:`input_shape` is undefined or non-square, or `rows` is not in [128, 160, 192, 224]. Weights for input shape (224, 224) will be loaded as the default.\n",
      "Number of layers in the pretrained mobilenet model:  86\n",
      "Model: \"sequential_9\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " time_distributed_12 (TimeDi  (None, 20, 3, 3, 1024)   3228864   \n",
      " stributed)                                                      \n",
      "                                                                 \n",
      " time_distributed_13 (TimeDi  (None, 20, 3, 3, 1024)   4096      \n",
      " stributed)                                                      \n",
      "                                                                 \n",
      " time_distributed_14 (TimeDi  (None, 20, 1, 1, 1024)   0         \n",
      " stributed)                                                      \n",
      "                                                                 \n",
      " time_distributed_15 (TimeDi  (None, 20, 1024)         0         \n",
      " stributed)                                                      \n",
      "                                                                 \n",
      " lstm_1 (LSTM)               (None, 128)               590336    \n",
      "                                                                 \n",
      " dropout_18 (Dropout)        (None, 128)               0         \n",
      "                                                                 \n",
      " dense_24 (Dense)            (None, 128)               16512     \n",
      "                                                                 \n",
      " dropout_19 (Dropout)        (None, 128)               0         \n",
      "                                                                 \n",
      " dense_25 (Dense)            (None, 5)                 645       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 3,840,453\n",
      "Trainable params: 3,816,517\n",
      "Non-trainable params: 23,936\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "num_epochs = 15 # choose the number of epochs\n",
    "print ('# epochs =', num_epochs)\n",
    "batch_size = 10  #experiment with the batch size\n",
    "print ('# batch size =', batch_size)\n",
    "num_frames = 20\n",
    "print ('# num frames per video =', num_frames)\n",
    "height = 120\n",
    "width = 120\n",
    "\n",
    "model = mobilenet_RNN(fine_tune=False, gru=False, num_images=num_frames)\n",
    "print(model.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "5Edy3nuDpCSf"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:`period` argument is deprecated. Please use `save_freq` to specify the frequency in number of batches seen.\n"
     ]
    }
   ],
   "source": [
    "callbacks_list = model_callbacks(\"model_retrain_mobilenet_lstm\")\n",
    "\n",
    "steps_per_epoch, validation_steps = calculate_steps(num_train_sequences, num_val_sequences, batch_size)\n",
    "\n",
    "train_generator = generate(train_path, train_doc, batch_size, num_images=num_frames)\n",
    "\n",
    "val_generator   = generate(val_path, val_doc, batch_size, num_images=num_frames)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "SWrZyY_vpCSh"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 1.1077 - categorical_accuracy: 0.5234\n",
      "Epoch 00001: val_loss improved from inf to 0.67545, saving model to model_retrain_mobilenet_lstm_2023-08-2907_28_23.340432/model-00001-1.10770-0.52338-0.67545-0.78000.h5\n",
      "67/67 [==============================] - 115s 2s/step - loss: 1.1077 - categorical_accuracy: 0.5234 - val_loss: 0.6754 - val_categorical_accuracy: 0.7800 - lr: 0.0010\n",
      "Epoch 2/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.4588 - categorical_accuracy: 0.8507\n",
      "Epoch 00002: val_loss improved from 0.67545 to 0.29834, saving model to model_retrain_mobilenet_lstm_2023-08-2907_28_23.340432/model-00002-0.45876-0.85068-0.29834-0.89000.h5\n",
      "67/67 [==============================] - 110s 2s/step - loss: 0.4588 - categorical_accuracy: 0.8507 - val_loss: 0.2983 - val_categorical_accuracy: 0.8900 - lr: 0.0010\n",
      "Epoch 3/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.3538 - categorical_accuracy: 0.8944\n",
      "Epoch 00003: val_loss did not improve from 0.29834\n",
      "67/67 [==============================] - 112s 2s/step - loss: 0.3538 - categorical_accuracy: 0.8944 - val_loss: 0.4512 - val_categorical_accuracy: 0.8500 - lr: 0.0010\n",
      "Epoch 4/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2895 - categorical_accuracy: 0.9125\n",
      "Epoch 00004: val_loss improved from 0.29834 to 0.10777, saving model to model_retrain_mobilenet_lstm_2023-08-2907_28_23.340432/model-00004-0.28946-0.91252-0.10777-0.96000.h5\n",
      "67/67 [==============================] - 115s 2s/step - loss: 0.2895 - categorical_accuracy: 0.9125 - val_loss: 0.1078 - val_categorical_accuracy: 0.9600 - lr: 0.0010\n",
      "Epoch 5/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2299 - categorical_accuracy: 0.9291\n",
      "Epoch 00005: val_loss did not improve from 0.10777\n",
      "67/67 [==============================] - 114s 2s/step - loss: 0.2299 - categorical_accuracy: 0.9291 - val_loss: 0.3219 - val_categorical_accuracy: 0.8900 - lr: 0.0010\n",
      "Epoch 6/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1670 - categorical_accuracy: 0.9412\n",
      "Epoch 00006: val_loss did not improve from 0.10777\n",
      "67/67 [==============================] - 107s 2s/step - loss: 0.1670 - categorical_accuracy: 0.9412 - val_loss: 0.1642 - val_categorical_accuracy: 0.9700 - lr: 0.0010\n",
      "Epoch 7/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.2097 - categorical_accuracy: 0.9367\n",
      "Epoch 00007: val_loss did not improve from 0.10777\n",
      "\n",
      "Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.\n",
      "67/67 [==============================] - 109s 2s/step - loss: 0.2097 - categorical_accuracy: 0.9367 - val_loss: 0.1145 - val_categorical_accuracy: 0.9400 - lr: 0.0010\n",
      "Epoch 8/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1170 - categorical_accuracy: 0.9668\n",
      "Epoch 00008: val_loss improved from 0.10777 to 0.09610, saving model to model_retrain_mobilenet_lstm_2023-08-2907_28_23.340432/model-00008-0.11704-0.96682-0.09610-0.96000.h5\n",
      "67/67 [==============================] - 109s 2s/step - loss: 0.1170 - categorical_accuracy: 0.9668 - val_loss: 0.0961 - val_categorical_accuracy: 0.9600 - lr: 1.0000e-04\n",
      "Epoch 9/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.1052 - categorical_accuracy: 0.9653\n",
      "Epoch 00009: val_loss did not improve from 0.09610\n",
      "67/67 [==============================] - 112s 2s/step - loss: 0.1052 - categorical_accuracy: 0.9653 - val_loss: 0.1264 - val_categorical_accuracy: 0.9500 - lr: 1.0000e-04\n",
      "Epoch 10/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0647 - categorical_accuracy: 0.9864\n",
      "Epoch 00010: val_loss improved from 0.09610 to 0.06719, saving model to model_retrain_mobilenet_lstm_2023-08-2907_28_23.340432/model-00010-0.06475-0.98643-0.06719-0.96000.h5\n",
      "67/67 [==============================] - 111s 2s/step - loss: 0.0647 - categorical_accuracy: 0.9864 - val_loss: 0.0672 - val_categorical_accuracy: 0.9600 - lr: 1.0000e-04\n",
      "Epoch 11/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0935 - categorical_accuracy: 0.9804\n",
      "Epoch 00011: val_loss did not improve from 0.06719\n",
      "67/67 [==============================] - 109s 2s/step - loss: 0.0935 - categorical_accuracy: 0.9804 - val_loss: 0.0805 - val_categorical_accuracy: 0.9600 - lr: 1.0000e-04\n",
      "Epoch 12/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0723 - categorical_accuracy: 0.9819\n",
      "Epoch 00012: val_loss improved from 0.06719 to 0.06227, saving model to model_retrain_mobilenet_lstm_2023-08-2907_28_23.340432/model-00012-0.07229-0.98190-0.06227-0.97000.h5\n",
      "67/67 [==============================] - 108s 2s/step - loss: 0.0723 - categorical_accuracy: 0.9819 - val_loss: 0.0623 - val_categorical_accuracy: 0.9700 - lr: 1.0000e-04\n",
      "Epoch 13/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0564 - categorical_accuracy: 0.9849\n",
      "Epoch 00013: val_loss improved from 0.06227 to 0.05053, saving model to model_retrain_mobilenet_lstm_2023-08-2907_28_23.340432/model-00013-0.05637-0.98492-0.05053-0.97000.h5\n",
      "67/67 [==============================] - 107s 2s/step - loss: 0.0564 - categorical_accuracy: 0.9849 - val_loss: 0.0505 - val_categorical_accuracy: 0.9700 - lr: 1.0000e-04\n",
      "Epoch 14/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0280 - categorical_accuracy: 0.9940\n",
      "Epoch 00014: val_loss did not improve from 0.05053\n",
      "67/67 [==============================] - 111s 2s/step - loss: 0.0280 - categorical_accuracy: 0.9940 - val_loss: 0.0532 - val_categorical_accuracy: 0.9700 - lr: 1.0000e-04\n",
      "Epoch 15/15\n",
      "67/67 [==============================] - ETA: 0s - loss: 0.0242 - categorical_accuracy: 0.9925\n",
      "Epoch 00015: val_loss improved from 0.05053 to 0.01912, saving model to model_retrain_mobilenet_lstm_2023-08-2907_28_23.340432/model-00015-0.02421-0.99246-0.01912-1.00000.h5\n",
      "67/67 [==============================] - 108s 2s/step - loss: 0.0242 - categorical_accuracy: 0.9925 - val_loss: 0.0191 - val_categorical_accuracy: 1.0000 - lr: 1.0000e-04\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(train_generator, steps_per_epoch=steps_per_epoch, epochs=num_epochs, verbose=1, \n",
    "                              callbacks=callbacks_list, validation_data=val_generator, \n",
    "                              validation_steps=validation_steps, class_weight=None, workers=1, initial_epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "VAhQsePBpCSj"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['loss', 'categorical_accuracy', 'val_loss', 'val_categorical_accuracy', 'lr'])\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_loss_accuracy(history)"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "Gesture_Recognition.ipynb",
   "provenance": [],
   "toc_visible": true
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}