Small change

src/cyclegan.py

@@ -5,12 +5,11 @@
 import os
 import time
 import multiprocessing
-import glob
 
 import tensorflow as tf
 
 import models
-from pipeline.load_data import load_train_data
+from pipeline.data import load_train_data
 from models.losses import generator_loss, discriminator_loss, cycle_consistency_loss, identity_loss
 from models.networks import Generator, Discriminator
 from utils.image_history_buffer import ImageHistoryBuffer
@@ -19,17 +18,18 @@
 
 """Hyperparameters (TODO: Move to argparse)"""
 project_dir = os.path.abspath(os.path.join(os.getcwd(), os.pardir))
-dataset_id = 'horse2zebra'
+dataset_id = 'facades'
 initial_learning_rate = 0.0002
-num_gen_filters = 32
+num_gen_filters = 64
 num_disc_filters = 64
 batch_size = 1 # Set batch size to 4 or 16 if training multigpu
 img_size = 256
 cyc_lambda = 10
 identity_lambda = 0
 if dataset_id == 'monet2photo':
  identity_lambda = 0.5
-epochs = 2
+epochs = 50
+save_epoch_freq = 5
 batches_per_epoch = models.get_batches_per_epoch(dataset_id, project_dir)
 
 def define_checkpoint(checkpoint_dir, model, training=True):
@@ -181,7 +181,7 @@ def train(data, model, checkpoint_info, epochs):
  # Assign decayed learning rate:
  learning_rate.assign(models.get_learning_rate(initial_learning_rate, global_step, batches_per_epoch))
  # Checkpoint the model:
- if (epoch + 1) % 5 == 0:
+ if (epoch + 1) % save_epoch_freq == 0:
  checkpoint_path = checkpoint.save(file_prefix=checkpoint_prefix)
  print("Checkpoint saved at ", checkpoint_path)
  print("Global Training Step: ", global_step.numpy() // 4)

src/pytorchtest.py

@@ -5,11 +5,13 @@
 import numpy as np
 
 """Pytorch testing ground"""
-dataset_id = 'horse2zebra'
+dataset_id = 'facades'
 project_dir = os.path.abspath(os.path.join(os.getcwd(), os.pardir))
 path_to_dataset = os.path.join(project_dir, 'data', 'raw', dataset_id + os.sep)
 testA_path = os.path.join(path_to_dataset, 'testA')
 img_path = testA_path + os.sep + 'n02381460_20.jpg'
 
 A = Image.open(img_path).convert('RGB')
 A = transforms.ToTensor()(A)
+
+# Things to try: ngf = 64, cyc_lambda = 1, 5, 20, identity loss, don't divide D loss by 2

src/test.py

@@ -8,11 +8,11 @@
 import tensorflow as tf
 
 from cyclegan import define_checkpoint, define_model, restore_from_checkpoint
-from pipeline.load_data import load_test_data, save_images
+from pipeline.data import load_test_data, save_images
 
 project_dir = os.path.abspath(os.path.join(os.getcwd(), os.pardir))
 checkpoint_dir = os.path.join(project_dir, 'saved_models', 'checkpoints')
-dataset_id = 'horse2zebra'
+dataset_id = 'facades'
 initial_learning_rate = 0.0002
 
 def test(data, model, checkpoint_info, dataset_id):
@@ -21,7 +21,7 @@ def test(data, model, checkpoint_info, dataset_id):
  generatedB = os.path.join(path_to_dataset, 'generatedB' + os.sep)
  genA2B = model['genA2B']
  genB2A = model['genB2A']
-
+ return None
  checkpoint, checkpoint_dir = checkpoint_info
  restore_from_checkpoint(checkpoint, checkpoint_dir)
  test_datasetA, test_datasetB, testA_size, testB_size = data
@@ -57,7 +57,7 @@ def test(data, model, checkpoint_info, dataset_id):
 if __name__ == "__main__":
  with tf.device("/cpu:0"): # Preprocess data on CPU for significant performance gains.
  data = load_test_data(dataset_id, project_dir)
- with tf.device("/gpu:0"):
+ #with tf.device("/gpu:0"):
  model = define_model(initial_learning_rate, training=False)
  checkpoint_info = define_checkpoint(checkpoint_dir, model, training=False)
  test(data, model, checkpoint_info, dataset_id)