Weakly Supervised Temporal Action Localization via Representative Snippet Knowledge Propagation

Weakly supervised temporal action localization aims to localize temporal boundaries of actions and simultaneously identify their categories with only video-level category labels. Many existing methods seek to generate pseudo labels for bridging the discrepancy between classification and localization, but usually only make use of limited contextual information for pseudo label generation. To alleviate this problem, we propose a representative snippet summarization and propagation framework. Our method seeks to mine the representative snippets in each video for propagating information between video snippets to generate better pseudo labels. For each video, its own representative snippets and the representative snippets from a memory bank are propagated to update the input features in an introand inter-video manner. The pseudo labels are generated from the temporal class activation maps of the updated features to rectify the predictions of the main branch. Our method obtains superior performance in comparison to the existing methods on two benchmarks, THUMOS14 and ActivityNet1.3, achieving gains as high as 1.2% in terms of average mAP on THUMOS14. Our code is available at https://github.com/LeonHLJ/RSKP.

Automated summary

Weakly supervised temporal action localization aims to localize the temporal boundaries of actions and identify their categories using only video-level labels. Existing methods often generate limited pseudo labels, but our proposed representative snippet summarization and propagation framework improves this by mining representation snippets and propagating information. Our method achieves superior performance on two benchmarks and outperforms existing methods, with gains as high as 1.2% in terms of average mAP on THUMOS14.