{"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/enhancing-person-re-identification-in-a-self","title":"Enhancing Person Re-identification in a Self-trained Subspace","arxiv_id":"1704.06020","date":"2017-04-20","proceeding":null,"authors":["Xun Yang","Meng Wang","Richang Hong","Qi Tian","Yong Rui"],"abstract":"Despite the promising progress made in recent years, person re-identification\n(re-ID) remains a challenging task due to the complex variations in human\nappearances from different camera views. For this challenging problem, a large\nvariety of algorithms have been developed in the fully-supervised setting,\nrequiring access to a large amount of labeled training data. However, the main\nbottleneck for fully-supervised re-ID is the limited availability of labeled\ntraining samples. To address this problem, in this paper, we propose a\nself-trained subspace learning paradigm for person re-ID which effectively\nutilizes both labeled and unlabeled data to learn a discriminative subspace\nwhere person images across disjoint camera views can be easily matched. The\nproposed approach first constructs pseudo pairwise relationships among\nunlabeled persons using the k-nearest neighbors algorithm. Then, with the\npseudo pairwise relationships, the unlabeled samples can be easily combined\nwith the labeled samples to learn a discriminative projection by solving an\neigenvalue problem. In addition, we refine the pseudo pairwise relationships\niteratively, which further improves the learning performance. A multi-kernel\nembedding strategy is also incorporated into the proposed approach to cope with\nthe non-linearity in person's appearance and explore the complementation of\nmultiple kernels. In this way, the performance of person re-ID can be greatly\nenhanced when training data are insufficient. Experimental results on six\nwidely-used datasets demonstrate the effectiveness of our approach and its\nperformance can be comparable to the reported results of most state-of-the-art\nfully-supervised methods while using much fewer labeled data.","url_abs":"http://arxiv.org/abs/1704.06020v2","url_pdf":"http://arxiv.org/pdf/1704.06020v2.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":"enhancing-person-re-identification-in-a-self","repo_url":"https://github.com/Xun-Yang/ReID_slef-training_TOMM2017","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"person-re-identification","task_name":"Person Re-Identification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}