This is the repository for BMVC 2017 paper "Cascaded Boundary Regression for Temporal Action Detection". If you find this helps your research, please cite
@inproceedings{gao2017cascaded,
title={Cascaded Boundary Regression for Temporal Action Detection},
author={Gao, Jiyang and Yang, Zhenheng and Nevatia, Ram},
booktitle={BMVC},
year={2017}
}
The model is built using TensorFlow 0.12, dataset is THUMOS-14.
In this work, two-stream CNN features are used. The appearance features can be downloaded at here: val set, test set; the denseflow features can be downloaded here: val set, test set. Note that, val set is used for training, as the train set for THUMOS-14 does not contain untrimmed videos.
The test action proposals are provided in test_proposals_from_TURN.txt. If you want to generate your own proposals, please go to TURN repository.
Modify the feature paths in main.py and run python main.py. The test results are saved to ./eval/test_results/. You need to run post-processing program to get the final detection results, python postproc.py xxx.pkl. The final detection results can be evaluated by THUMOS-14 official eval tool.
I uploaded my detection results from tIoU=0.1 to tIoU=0.7 here