Source code for Finding Action Tubes, created by Georgia Gkioxari at UC Berkeley.

1. Requirements
2. Usage
3. Instructions
4. Downloads

You need Caffe. Please refer to Caffe installation instructions

The pipeline consists of multiple steps. For simplicity, we break them down to independent procedures.

1. In startup.m, add the paths and resolve the dependencies

2. Selective search boxes need to be stored in the following format
   /ss_dir/motion/action/video/0000f.mat
   Video frames need to be stored in the following format
   /img_dir/action/video/0000f.png

3. Optical flow computation

    compute_OF/compute_flow.m  
    	for a pair of images computes flow as described in the paper.  
    		im1, im2: input images  
    		optical flow images need to be stored in the format `/flow_dir/action/video/0000f.png`
   

4. Motion saliency

    motion_saliency/get_motion_salient_boxes.m  
    	for each frame, prunes boxes (e.g. from Selective Search) based the optical flow signal within each box.  
    		 annot: set of videos and actions, jhmdb_annot.mat  
    		 ss_dir: directory containing the boxes  
    		 flow_dir: directory containing the optical flow images (as computed by A.)
   

5. Extract fc7 features

    extract_features/rcnn_cache_fc7_features_jhmdb.m  
    	extracts fc7 features  
    		 type: 'spatial' or 'motion'  
    		 net_def_file: prototxts, `models/jhmdb/extract_fc7.prototxt`. Same for any type  
    		 net_file: models, use pretrained models for JHMDB as provided in the project page  
    		 output_dir: cache directory, the features are cached in output_dir/type/action/video/frame.mat
   

6. Train SVM models

    train_svm/train_jhmdb.m  
    	trains SVM models, one for each action  
    		annot: ground truth information and boxes (after pruning), jhmdb_motion_sal_annot.mat  
    		feat_dir: directory with cached features  
    		save_dir: cache directory
   

7. Action Tubes

    train_svm/compute_tubes.m  
    	scores and links detections to create the final action tubes  
    		annot: source of boxes, `jhmdb_motion_sal_annot.mat`  
    		rcnn_model: the models as computed by train_jhmdb.m
   

8. Precomputed tubes

    test_tubes/  
    	tubes for all three splits of J-HMDB and UCF sports  
    test_tubes/UCFsports_benchmark/  
    	AUC and ROC numbers for UCFSports and plots (see ipython notebook)
   

9. Evaluate/ROC curves

    evaluate/get_ROC_curve_JHMDB.m  
    	computes ROC and AUC for JHMDB  
    		annot: ground truth annotation (annot_jhdmb.mat)  
    		tubes: tubes on the test set  
    		actions: list of actions  
    		iou_thresh: threshold for intersection over union  
    		draw: true to draw the curves  
    (For UCF sports the same function was used with some small adjustments regarding the format of the data)
   

To train action tubes from scratch, you need to train two networks.

Training spatial-CNN and motion-CNN

1. Compute the optical flow as in A.
2. Create `window_train(val).txt` with the window data (similar to R-CNN detection)
3. Use Caffe to train (train prototxt is given in `models/jhmdb/train.prototxt`), and initialize with the proper model
4. In the case of motion-CNN, you need to make two changes in `window_data_layer.cpp`
	a. The image mean is for all channels 128 (instead of the image mean provided)
	b. During training, when flipping of the input image occurs the flow in x needs to also change sign

Download pretrained models from the project's webpage