{"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/tracking-emerges-by-colorizing-videos","title":"Tracking Emerges by Colorizing Videos","arxiv_id":"1806.09594","date":"2018-06-25","proceeding":"ECCV 2018 9","authors":["Carl Vondrick","Abhinav Shrivastava","Alireza Fathi","Sergio Guadarrama","Kevin Murphy"],"abstract":"We use large amounts of unlabeled video to learn models for visual tracking\nwithout manual human supervision. We leverage the natural temporal coherency of\ncolor to create a model that learns to colorize gray-scale videos by copying\ncolors from a reference frame. Quantitative and qualitative experiments suggest\nthat this task causes the model to automatically learn to track visual regions.\nAlthough the model is trained without any ground-truth labels, our method\nlearns to track well enough to outperform the latest methods based on optical\nflow. Moreover, our results suggest that failures to track are correlated with\nfailures to colorize, indicating that advancing video colorization may further\nimprove self-supervised visual tracking.","url_abs":"http://arxiv.org/abs/1806.09594v2","url_pdf":"http://arxiv.org/pdf/1806.09594v2.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":"tracking-emerges-by-colorizing-videos","repo_url":"https://github.com/hyperparameters/tracking_via_colorization","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"colorization","task_name":"Colorization"},{"task_slug":"optical-flow-estimation","task_name":"Optical Flow Estimation"},{"task_slug":"skeleton-based-action-recognition","task_name":"Skeleton Based Action Recognition"},{"task_slug":"visual-tracking","task_name":"Visual Tracking"}],"methods":[{"method_slug":"colorization","method_name":"Colorization"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/skeleton-based-action-recognition-on-jhmdb","task":"Skeleton Based Action Recognition","dataset":"JHMDB Pose Tracking","model":"ColorPointer","rank_in_archive_order":2,"of":3,"metrics":{"PCK@0.1":"45.2","PCK@0.2":"69.6","PCK@0.3":"80.8","PCK@0.4":"87.5","PCK@0.5":"91.4"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1806.09594","atlas_url":"https://app.syntology.ai/?focus=1806.09594","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}