{"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/a-pursuit-of-temporal-accuracy-in-general","title":"A Pursuit of Temporal Accuracy in General Activity Detection","arxiv_id":"1703.02716","date":"2017-03-08","proceeding":null,"authors":["Yuanjun Xiong","Yue Zhao","Li-Min Wang","Dahua Lin","Xiaoou Tang"],"abstract":"Detecting activities in untrimmed videos is an important but challenging\ntask. The performance of existing methods remains unsatisfactory, e.g., they\noften meet difficulties in locating the beginning and end of a long complex\naction. In this paper, we propose a generic framework that can accurately\ndetect a wide variety of activities from untrimmed videos. Our first\ncontribution is a novel proposal scheme that can efficiently generate\ncandidates with accurate temporal boundaries. The other contribution is a\ncascaded classification pipeline that explicitly distinguishes between\nrelevance and completeness of a candidate instance. On two challenging temporal\nactivity detection datasets, THUMOS14 and ActivityNet, the proposed framework\nsignificantly outperforms the existing state-of-the-art methods, demonstrating\nsuperior accuracy and strong adaptivity in handling activities with various\ntemporal structures.","url_abs":"http://arxiv.org/abs/1703.02716v1","url_pdf":"http://arxiv.org/pdf/1703.02716v1.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":[{"paper_slug":"a-pursuit-of-temporal-accuracy-in-general","repo_url":"https://github.com/yjxiong/action-detection","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"action-detection","task_name":"Action Detection"},{"task_slug":"activity-detection","task_name":"Activity Detection"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"action-recognition","task_name":"Temporal Action Localization"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/temporal-action-localization-on-activitynet","task":"Temporal Action Localization","dataset":"ActivityNet-1.3","model":"SSN","rank_in_archive_order":30,"of":33,"metrics":{"mAP":"32.26","mAP IOU@0.5":"39.12"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1703.02716","atlas_url":"https://app.syntology.ai/?focus=1703.02716","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}