{"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/adaptive-correlation-filters-with-long-term","title":"Adaptive Correlation Filters with Long-Term and Short-Term Memory for Object Tracking","arxiv_id":"1707.02309","date":"2017-07-07","proceeding":null,"authors":["Chao Ma","Jia-Bin Huang","Xiaokang Yang","Ming-Hsuan Yang"],"abstract":"Object tracking is challenging as target objects often undergo drastic\nappearance changes over time. Recently, adaptive correlation filters have been\nsuccessfully applied to object tracking. However, tracking algorithms relying\non highly adaptive correlation filters are prone to drift due to noisy updates.\nMoreover, as these algorithms do not maintain long-term memory of target\nappearance, they cannot recover from tracking failures caused by heavy\nocclusion or target disappearance in the camera view. In this paper, we propose\nto learn multiple adaptive correlation filters with both long-term and\nshort-term memory of target appearance for robust object tracking. First, we\nlearn a kernelized correlation filter with an aggressive learning rate for\nlocating target objects precisely. We take into account the appropriate size of\nsurrounding context and the feature representations. Second, we learn a\ncorrelation filter over a feature pyramid centered at the estimated target\nposition for predicting scale changes. Third, we learn a complementary\ncorrelation filter with a conservative learning rate to maintain long-term\nmemory of target appearance. We use the output responses of this long-term\nfilter to determine if tracking failure occurs. In the case of tracking\nfailures, we apply an incrementally learned detector to recover the target\nposition in a sliding window fashion. Extensive experimental results on\nlarge-scale benchmark datasets demonstrate that the proposed algorithm performs\nfavorably against the state-of-the-art methods in terms of efficiency,\naccuracy, and robustness.","url_abs":"http://arxiv.org/abs/1707.02309v2","url_pdf":"http://arxiv.org/pdf/1707.02309v2.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":"adaptive-correlation-filters-with-long-term","repo_url":"https://github.com/chaoma99/lct-tracker","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"object-tracking","task_name":"Object Tracking"},{"task_slug":null,"task_name":"Position"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}