Tensorflow Implementation of the Paper SST: Single-Stream Temporal Action Proposals by Shyamal Buch et al. in CVPR 2017.
Please download video data and annotation data from the website THUMOS14. Extract C3D features for non-overlap 16-frame snippets from the 412 videos (200 val videos + 212 test videos, I found one test video missing) for the task of temporal action proposals. Alternatively, you may download my provided C3D features, and put them in dataset/thumos14/features/. If you are interested in the feature extraction, I refer you to this code.
fc6 features are used in my experiment.
Please follow the script dataset/thumos14/prepare_gt_proposal_data.py to generate ground-truth proposal data for train/val/test split. I already put the generated files in dataset/thumos14/gt_proposals/.
After that, please generate anchor weights (for handling imbalance class problem) by uniformly sampling video streams (follow dataset/thumos14/anchors/get_anchor_weight.py) or just use my pre-calculated weights (weights.json).
The best configuration (from my experiments) is given in opt.py, including model setup, training options, and testing options.
Train your model using the script train.py. Run around 50 epochs and pick the best checkpoint (with the smallest val loss) for prediction.
Follow the script test.py to make proposal predictions.
Follow the script eval.py to evaluate your proposal predictions.
You may download my trained model here. Please put them in checkpoints/. Change the file init_from in opt.py and run test.py !
Update: The predicted action proposals for the test set can be found here. The result figures are put in results/1. They are slightly better than the reported ones.
| Method | Recall@1000 at tIoU=0.8 |
|---|---|
| SST (paper) | 0.672 |
| SST (my impl) | 0.696 |
tensorflow==1.0.1
python==2.7.5
Other versions may also work.
Great thanks to Shyamal Buch for really helpful discussion.