{"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/pose-flow-efficient-online-pose-tracking","title":"Pose Flow: Efficient Online Pose Tracking","arxiv_id":"1802.00977","date":"2018-02-03","proceeding":null,"authors":["Yuliang Xiu","Jiefeng Li","Haoyu Wang","Yinghong Fang","Cewu Lu"],"abstract":"Multi-person articulated pose tracking in unconstrained videos is an\nimportant while challenging problem. In this paper, going along the road of\ntop-down approaches, we propose a decent and efficient pose tracker based on\npose flows. First, we design an online optimization framework to build the\nassociation of cross-frame poses and form pose flows (PF-Builder). Second, a\nnovel pose flow non-maximum suppression (PF-NMS) is designed to robustly reduce\nredundant pose flows and re-link temporal disjoint ones. Extensive experiments\nshow that our method significantly outperforms best-reported results on two\nstandard Pose Tracking datasets by 13 mAP 25 MOTA and 6 mAP 3 MOTA\nrespectively. Moreover, in the case of working on detected poses in individual\nframes, the extra computation of pose tracker is very minor, guaranteeing\nonline 10FPS tracking. Our source codes are made publicly\navailable(https://github.com/YuliangXiu/PoseFlow).","url_abs":"http://arxiv.org/abs/1802.00977v2","url_pdf":"http://arxiv.org/pdf/1802.00977v2.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":"pose-flow-efficient-online-pose-tracking","repo_url":"https://github.com/YuliangXiu/PoseFlow","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"pose-tracking","task_name":"Pose Tracking"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/keypoint-detection-on-coco-test-challenge","task":"Keypoint Detection","dataset":"COCO test-challenge","model":"Xiu et al.","rank_in_archive_order":8,"of":8,"metrics":{"AR":"67.5","ARM":"62.5"},"uses_additional_data":false},{"leaderboard":"/sota/pose-tracking-on-posetrack2017","task":"Pose Tracking","dataset":"PoseTrack2017","model":"PoseFlow","rank_in_archive_order":9,"of":10,"metrics":{"MOTA":"50.98","mAP":"62.95"},"uses_additional_data":true}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1802.00977","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}