Rethinking the Faster R-CNN Architecture for Temporal Action Localization

Abstract

We propose TAL-Net, an improved approach to temporal action localization in video that is inspired by the Faster R-CNN object detection framework. TAL-Net addresses three key shortcomings of existing approaches: (1) we improve receptive field alignment using a multi-scale architecture that can accommodate extreme variation in action durations; (2) we better exploit the temporal context of actions for both proposal generation and action classification by appropriately extending receptive fields; and (3) we explicitly consider multi-stream feature fusion and demonstrate that fusing motion late is important. We achieve state-of-the-art performance for both action proposal and localization on THUMOS'14 detection benchmark and competitive performance on ActivityNet challenge.

Cite

Text

Chao et al. "Rethinking the Faster R-CNN Architecture for Temporal Action Localization." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018. doi:10.1109/CVPR.2018.00124

Markdown

[Chao et al. "Rethinking the Faster R-CNN Architecture for Temporal Action Localization." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018.](https://mlanthology.org/cvpr/2018/chao2018cvpr-rethinking/) doi:10.1109/CVPR.2018.00124

BibTeX

@inproceedings{chao2018cvpr-rethinking,
  title     = {{Rethinking the Faster R-CNN Architecture for Temporal Action Localization}},
  author    = {Chao, Yu-Wei and Vijayanarasimhan, Sudheendra and Seybold, Bryan and Ross, David A. and Deng, Jia and Sukthankar, Rahul},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year      = {2018},
  doi       = {10.1109/CVPR.2018.00124},
  url       = {https://mlanthology.org/cvpr/2018/chao2018cvpr-rethinking/}
}