{"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/real-time-visual-tracking-by-deep-reinforced","title":"Real-time visual tracking by deep reinforced decision making","arxiv_id":"1702.06291","date":"2017-02-21","proceeding":null,"authors":["Janghoon Choi","Junseok Kwon","Kyoung Mu Lee"],"abstract":"One of the major challenges of model-free visual tracking problem has been\nthe difficulty originating from the unpredictable and drastic changes in the\nappearance of objects we target to track. Existing methods tackle this problem\nby updating the appearance model on-line in order to adapt to the changes in\nthe appearance. Despite the success of these methods however, inaccurate and\nerroneous updates of the appearance model result in a tracker drift. In this\npaper, we introduce a novel real-time visual tracking algorithm based on a\ntemplate selection strategy constructed by deep reinforcement learning methods.\nThe tracking algorithm utilizes this strategy to choose the appropriate\ntemplate for tracking a given frame. The template selection strategy is\nself-learned by utilizing a simple policy gradient method on numerous training\nepisodes randomly generated from a tracking benchmark dataset. Our proposed\nreinforcement learning framework is generally applicable to other confidence\nmap based tracking algorithms. The experiment shows that our tracking algorithm\nruns in real-time speed of 43 fps and the proposed policy network effectively\ndecides the appropriate template for successful visual tracking.","url_abs":"http://arxiv.org/abs/1702.06291v2","url_pdf":"http://arxiv.org/pdf/1702.06291v2.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":"real-time-visual-tracking-by-deep-reinforced","repo_url":"https://github.com/JanghoonChoi/janghoonchoi.github.io","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"deep-reinforcement-learning","task_name":"Deep Reinforcement Learning"},{"task_slug":"real-time-visual-tracking","task_name":"Real-Time Visual Tracking"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"},{"task_slug":"visual-tracking","task_name":"Visual Tracking"},{"task_slug":"reinforcement-learning-2","task_name":"reinforcement-learning"}],"methods":[{"method_slug":"speed","method_name":"SPEED"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}