{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/pat-position-aware-transformer-for-dense","title":"PAT: Position-Aware Transformer for Dense Multi-Label Action Detection","arxiv_id":"2308.05051","date":"2023-08-09","proceeding":null,"authors":["Faegheh Sardari","Armin Mustafa","Philip J. B. Jackson","Adrian Hilton"],"abstract":"We present PAT, a transformer-based network that learns complex temporal co-occurrence action dependencies in a video by exploiting multi-scale temporal features. In existing methods, the self-attention mechanism in transformers loses the temporal positional information, which is essential for robust action detection. To address this issue, we (i) embed relative positional encoding in the self-attention mechanism and (ii) exploit multi-scale temporal relationships by designing a novel non hierarchical network, in contrast to the recent transformer-based approaches that use a hierarchical structure. We argue that joining the self-attention mechanism with multiple sub-sampling processes in the hierarchical approaches results in increased loss of positional information. We evaluate the performance of our proposed approach on two challenging dense multi-label benchmark datasets, and show that PAT improves the current state-of-the-art result by 1.1% and 0.6% mAP on the Charades and MultiTHUMOS datasets, respectively, thereby achieving the new state-of-the-art mAP at 26.5% and 44.6%, respectively. We also perform extensive ablation studies to examine the impact of the different components of our proposed network.","url_abs":"https://arxiv.org/abs/2308.05051v1","url_pdf":"https://arxiv.org/pdf/2308.05051v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[],"tasks":[{"task_slug":"action-detection","task_name":"Action Detection"},{"task_slug":"event-detection","task_name":"Event Detection"},{"task_slug":null,"task_name":"Position"}],"methods":[{"method_slug":"non","method_name":"NON"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/action-detection-on-charades","task":"Action Detection","dataset":"Charades","model":"PAT","rank_in_archive_order":6,"of":16,"metrics":{"mAP":"26.5"},"uses_additional_data":false},{"leaderboard":"/sota/action-detection-on-multithumos-1","task":"Action Detection","dataset":"MultiTHUMOS","model":"PAT","rank_in_archive_order":1,"of":1,"metrics":{"mAP":"44.6"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2308.05051","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}