Sub-action Prototype Learning for Point-level Weakly-supervised Temporal Action Localization
This work addresses a domain-specific challenge in video analysis for applications like surveillance or sports, offering an incremental improvement over existing methods.
The paper tackles the problem of point-level weakly-supervised temporal action localization, where only single timestamps are available, by proposing a sub-action prototype learning framework that improves boundary prediction, achieving significant performance gains over state-of-the-art methods on three benchmarks.
Point-level weakly-supervised temporal action localization (PWTAL) aims to localize actions with only a single timestamp annotation for each action instance. Existing methods tend to mine dense pseudo labels to alleviate the label sparsity, but overlook the potential sub-action temporal structures, resulting in inferior performance. To tackle this problem, we propose a novel sub-action prototype learning framework (SPL-Loc) which comprises Sub-action Prototype Clustering (SPC) and Ordered Prototype Alignment (OPA). SPC adaptively extracts representative sub-action prototypes which are capable to perceive the temporal scale and spatial content variation of action instances. OPA selects relevant prototypes to provide completeness clue for pseudo label generation by applying a temporal alignment loss. As a result, pseudo labels are derived from alignment results to improve action boundary prediction. Extensive experiments on three popular benchmarks demonstrate that the proposed SPL-Loc significantly outperforms existing SOTA PWTAL methods.