{"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/learning-to-update-for-object-tracking-with","title":"Learning to Update for Object Tracking with Recurrent Meta-learner","arxiv_id":"1806.07078","date":"2018-06-19","proceeding":null,"authors":["Bi Li","Wenxuan Xie","Wen-Jun Zeng","Wenyu Liu"],"abstract":"Model update lies at the heart of object tracking. Generally, model update is\nformulated as an online learning problem where a target model is learned over\nthe online training set. Our key innovation is to \\emph{formulate the model\nupdate problem in the meta-learning framework and learn the online learning\nalgorithm itself using large numbers of offline videos}, i.e., \\emph{learning\nto update}. The learned updater takes as input the online training set and\noutputs an updated target model. As a first attempt, we design the learned\nupdater based on recurrent neural networks (RNNs) and demonstrate its\napplication in a template-based tracker and a correlation filter-based tracker.\nOur learned updater consistently improves the base trackers and runs faster\nthan realtime on GPU while requiring small memory footprint during testing.\nExperiments on standard benchmarks demonstrate that our learned updater\noutperforms commonly used update baselines including the efficient exponential\nmoving average (EMA)-based update and the well-designed stochastic gradient\ndescent (SGD)-based update. Equipped with our learned updater, the\ntemplate-based tracker achieves state-of-the-art performance among realtime\ntrackers on GPU.","url_abs":"http://arxiv.org/abs/1806.07078v3","url_pdf":"http://arxiv.org/pdf/1806.07078v3.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":[],"tasks":[{"task_slug":null,"task_name":"GPU"},{"task_slug":"meta-learning","task_name":"Meta-Learning"},{"task_slug":"object-tracking","task_name":"Object Tracking"},{"task_slug":"visual-object-tracking","task_name":"Visual Object Tracking"},{"task_slug":"visual-tracking","task_name":"Visual Tracking"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/visual-object-tracking-on-vot2016","task":"Visual Object Tracking","dataset":"VOT2016","model":"SiamFC-lu (Ours)","rank_in_archive_order":6,"of":6,"metrics":{"Expected Average Overlap (EAO)":"0.295"},"uses_additional_data":false},{"leaderboard":"/sota/visual-object-tracking-on-vot2017","task":"Visual Object Tracking","dataset":"VOT2017","model":"SiamFC-lu (Ours)","rank_in_archive_order":5,"of":6,"metrics":{"Expected Average Overlap (EAO)":"0.263"},"uses_additional_data":false},{"leaderboard":"/sota/visual-tracking-on-otb-100","task":"Visual Tracking","dataset":"OTB-100","model":"SiamFC-lu (Ours)","rank_in_archive_order":1,"of":1,"metrics":{"AUC":"0.318"},"uses_additional_data":false},{"leaderboard":"/sota/visual-tracking-on-otb-2013","task":"Visual Tracking","dataset":"OTB-2013","model":"SiamFC-lu (Ours)","rank_in_archive_order":1,"of":1,"metrics":{"AUC":"0.657"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}