{"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/multiview-supervision-by-registration","title":"Multiview Supervision By Registration","arxiv_id":"1811.11251","date":"2018-11-27","proceeding":null,"authors":["Yilun Zhang","Hyun Soo Park"],"abstract":"This paper presents a semi-supervised learning framework to train a keypoint\ndetector using multiview image streams given the limited labeled data\n(typically $<$4\\%). We leverage the complementary relationship between\nmultiview geometry and visual tracking to provide three types of supervisionary\nsignals to utilize the unlabeled data: (1) keypoint detection in one view can\nbe supervised by other views via the epipolar geometry; (2) a keypoint moves\nsmoothly over time where its optical flow can be used to temporally supervise\nconsecutive image frames to each other; (3) visible keypoint in one view is\nlikely to be visible in the adjacent view. We integrate these three signals in\na differentiable fashion to design a new end-to-end neural network composed of\nthree pathways. This design allows us to extensively use the unlabeled data to\ntrain the keypoint detector. We show that our approach outperforms existing\ndetectors including DeepLabCut tailored to the keypoint detection of non-human\nspecies such as monkeys, dogs, and mice.","url_abs":"http://arxiv.org/abs/1811.11251v2","url_pdf":"http://arxiv.org/pdf/1811.11251v2.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":"multiview-supervision-by-registration","repo_url":"https://github.com/msbrpp/MSBR","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"3d-reconstruction","task_name":"3D Reconstruction"},{"task_slug":"keypoint-detection","task_name":"Keypoint Detection"},{"task_slug":"optical-flow-estimation","task_name":"Optical Flow Estimation"},{"task_slug":"visual-tracking","task_name":"Visual Tracking"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1811.11251","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}