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

Author: ehofesmann

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)"
     }
-]
+]