Update json template frame size, fix num_clip=0 bug (#3)

* Update json template frame size, fix num_clip=0 bug
  Update README.md with link to wiki

README.md

@@ -2,6 +2,8 @@
 
 A platform for quick and easy development of deep learning networks for recognition and detection in videos. Includes popular models like C3D and SSD.
 
+Check out our [wiki!](https://github.com/MichiganCOG/ViP/wiki)
+
 ## Implemented Models and their performance
 
 ### Recognition
@@ -26,7 +28,7 @@ A platform for quick and easy development of deep learning networks for recognit
 * [Development](#development)
  * [Add a Model](#add-a-model)
  * [Add a Dataset](#add-a-dataset)
-* [Version History](#version-history)
+* [FAQ](#faq)
 
 ## Configured Datasets
 | Dataset | Task(s) |
@@ -88,6 +90,9 @@ Ex: From the root directory of ViP, train the action recognition network C3D on
 ```
 python train.py --cfg_file models/c3d/config_train.yaml
 ```
+
+Additional examples can be found on our [wiki.](https://github.com/MichiganCOG/ViP/wiki)
+
 ## Development
 
 New models and datasets can be added without needing to rewrite any training, evaluation, or data loading code.
@@ -103,6 +108,8 @@ To add a new model:
 
 Examples of previously implemented models can be found [here](https://github.com/MichiganCOG/ViP/tree/master/models).
 
+Additional information can be found on our [wiki.](https://github.com/MichiganCOG/ViP/wiki)
+
 ### Add a Dataset
 
 To add a new dataset:
@@ -114,3 +121,8 @@ To add a new dataset:
  * Complete `__init__` and `__getitem__` functions
  * Example skeleton dataset can be found [here](https://github.com/MichiganCOG/ViP/blob/master/datasets/templates/dataset_template.py)
 
+Additional information can be found on our [wiki.](https://github.com/MichiganCOG/ViP/wiki)
+
+### FAQ
+
+A detailed FAQ can be found on our [wiki](https://github.com/MichiganCOG/ViP/wiki/FAQ).

datasets/abstract_datasets.py

@@ -71,6 +71,7 @@ def _extractClips(self, video):
  if self.num_clips < 0:
  if len(video) >= self.clip_length:
  final_video = [video[_idx] for _idx in np.linspace(0, len(video)-1, self.clip_length, dtype='int32')]
+ final_video = [final_video]
 
  else:
  # Loop if insufficient elements
@@ -80,6 +81,7 @@ def _extractClips(self, video):
  indices = indices[np.linspace(0, len(indices)-1, self.clip_length, dtype='int32')]
 
  final_video = [video[_idx] for _idx in indices]
+ final_video = [final_video]
 
 
  # END IF
@@ -103,6 +105,7 @@ def _extractClips(self, video):
  indices = indices[:self.clip_length]
 
  final_video = [video[_idx] for _idx in indices]
+ final_video = [final_video]
 
  # END IF 
 
@@ -114,6 +117,7 @@ def _extractClips(self, video):
  indices = np.arange(indices, indices + self.clip_length).astype('int32') 
 
  final_video = [video[_idx] for _idx in indices]
+ final_video = [final_video]
 
  else:
  indices = np.ceil(self.clip_length/float(len(video)))
@@ -125,17 +129,19 @@ def _extractClips(self, video):
  indices = indices[index:index + self.clip_length]
 
  final_video = [video[_idx] for _idx in indices]
+ final_video = [final_video]
 
  # END IF
 
  else:
  final_video = video[:self.clip_length]
+ final_video = [final_video]
 
  # END IF
 
  # END IF
 
- return [final_video]
+ return final_video
 
 
 

datasets/templates/action_recognition_template.json

@@ -2,7 +2,6 @@
  {
  "frames (list)": [
  {
- "frame_size (int, int)": "(WIDTH,HEIGHT)",
  "img_path (str)": "FRAME_PATH",
  "actions (list)": [
  {
@@ -11,6 +10,7 @@
  ]
  }
  ],
+ "frame_size (int, int)": "(WIDTH,HEIGHT)",
  "base_path (str)": "BASE_VID_PATH"
  }
-]
+]

datasets/templates/detection_template.json

@@ -2,7 +2,6 @@
  {
  "frames (list)": [
  {
- "frame_size (int, int)": "(WIDTH,HEIGHT)",
  "img_path (str)": "FRAME_PATH",
  "objs (list)": [
  {
@@ -14,6 +13,7 @@
  ]
  }
  ],
- "base_path (str)": "BASE_VID_PATH"
+ "base_path (str)": "BASE_VID_PATH",
+ "frame_size (int, int)": "(WIDTH,HEIGHT)"
  }
-]
+]