Real-Time Temporal Action Localization In Untrimmed Videos By Sub-Action Discovery
Abstract
This paper presents a computationally efficient approach for temporal action detection in untrimmed videos that outperforms state-of-the-art methods by a large margin. We exploit the temporal structure of actions by modeling an action as a sequence of sub-actions. A novel and fully automatic sub-action discovery algorithm is proposed, where the number of sub-actions for each action as well as their types are automatically determined from the training videos. We find that the discovered sub-actions are semantically meaningful. To localize an action, an objective function combining appearance, duration and temporal structure of sub-actions is optimized as a shortest path problem in a network flow formulation. A significant benefit of the proposed approach is that it enables real-time action localization (40 fps) in untrimmed videos. We demonstrate state-of-the-art results on THUMOS’14 and MEXaction2 datasets.
Publication Date
1-1-2017
Publication Title
British Machine Vision Conference 2017, BMVC 2017
Document Type
Article; Proceedings Paper
Personal Identifier
scopus
DOI Link
https://doi.org/10.5244/c.31.91
Copyright Status
Unknown
Socpus ID
85088771071 (Scopus)
Source API URL
https://api.elsevier.com/content/abstract/scopus_id/85088771071
STARS Citation
Hou, Rui; Sukthankar, Rahul; and Shah, Mubarak, "Real-Time Temporal Action Localization In Untrimmed Videos By Sub-Action Discovery" (2017). Scopus Export 2015-2019. 6700.
https://stars.library.ucf.edu/scopus2015/6700