{"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/discovering-underlying-person-structure","title":"Discovering Underlying Person Structure Pattern with Relative Local Distance for Person Re-identification","arxiv_id":"1901.10100","date":"2019-01-29","proceeding":null,"authors":["Guangcong Wang","Jian-Huang Lai","Zhenyu Xie","Xiaohua Xie"],"abstract":"Modeling the underlying person structure for person re-identification (re-ID)\nis difficult due to diverse deformable poses, changeable camera views and\nimperfect person detectors. How to exploit underlying person structure\ninformation without extra annotations to improve the performance of person\nre-ID remains largely unexplored. To address this problem, we propose a novel\nRelative Local Distance (RLD) method that integrates a relative local distance\nconstraint into convolutional neural networks (CNNs) in an end-to-end way. It\nis the first time that the relative local constraint is proposed to guide the\nglobal feature representation learning. Specially, a relative local distance\nmatrix is computed by using feature maps and then regarded as a regularizer to\nguide CNNs to learn a structure-aware feature representation. With the\ndiscovered underlying person structure, the RLD method builds a bridge between\nthe global and local feature representation and thus improves the capacity of\nfeature representation for person re-ID. Furthermore, RLD also significantly\naccelerates deep network training compared with conventional methods. The\nexperimental results show the effectiveness of RLD on the CUHK03, Market-1501,\nand DukeMTMC-reID datasets. Code is available at\n\\url{https://github.com/Wanggcong/RLD_codes}.","url_abs":"http://arxiv.org/abs/1901.10100v1","url_pdf":"http://arxiv.org/pdf/1901.10100v1.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":"discovering-underlying-person-structure","repo_url":"https://github.com/Wanggcong/RLD_codes","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"person-re-identification","task_name":"Person Re-Identification"},{"task_slug":"representation-learning","task_name":"Representation Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}