{"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-by-animation-unsupervised-learning","title":"Tracking by Animation: Unsupervised Learning of Multi-Object Attentive Trackers","arxiv_id":"1809.03137","date":"2018-09-10","proceeding":"CVPR 2019 6","authors":["Zhen He","Jian Li","Daxue Liu","Hangen He","David Barber"],"abstract":"Online Multi-Object Tracking (MOT) from videos is a challenging computer\nvision task which has been extensively studied for decades. Most of the\nexisting MOT algorithms are based on the Tracking-by-Detection (TBD) paradigm\ncombined with popular machine learning approaches which largely reduce the\nhuman effort to tune algorithm parameters. However, the commonly used\nsupervised learning approaches require the labeled data (e.g., bounding boxes),\nwhich is expensive for videos. Also, the TBD framework is usually suboptimal\nsince it is not end-to-end, i.e., it considers the task as detection and\ntracking, but not jointly. To achieve both label-free and end-to-end learning\nof MOT, we propose a Tracking-by-Animation framework, where a differentiable\nneural model first tracks objects from input frames and then animates these\nobjects into reconstructed frames. Learning is then driven by the\nreconstruction error through backpropagation. We further propose a\nReprioritized Attentive Tracking to improve the robustness of data association.\nExperiments conducted on both synthetic and real video datasets show the\npotential of the proposed model. Our project page is publicly available at:\nhttps://github.com/zhen-he/tracking-by-animation","url_abs":"http://arxiv.org/abs/1809.03137v3","url_pdf":"http://arxiv.org/pdf/1809.03137v3.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-by-animation-unsupervised-learning","repo_url":"https://github.com/zhen-he/tracking-by-animation","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"multi-object-tracking","task_name":"Multi-Object Tracking"},{"task_slug":"object-tracking","task_name":"Object Tracking"},{"task_slug":"online-multi-object-tracking","task_name":"Online Multi-Object Tracking"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1809.03137","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}