{"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/unsupervised-person-re-identification-by-deep","title":"Unsupervised Person Re-identification by Deep Asymmetric Metric Embedding","arxiv_id":"1901.10177","date":"2019-01-29","proceeding":null,"authors":["Hong-Xing Yu","An-Cong Wu","Wei-Shi Zheng"],"abstract":"Person re-identification (Re-ID) aims to match identities across\nnon-overlapping camera views. Researchers have proposed many supervised Re-ID\nmodels which require quantities of cross-view pairwise labelled data. This\nlimits their scalabilities to many applications where a large amount of data\nfrom multiple disjoint camera views is available but unlabelled. Although some\nunsupervised Re-ID models have been proposed to address the scalability\nproblem, they often suffer from the view-specific bias problem which is caused\nby dramatic variances across different camera views, e.g., different\nillumination, viewpoints and occlusion. The dramatic variances induce specific\nfeature distortions in different camera views, which can be very disturbing in\nfinding cross-view discriminative information for Re-ID in the unsupervised\nscenarios, since no label information is available to help alleviate the bias.\nWe propose to explicitly address this problem by learning an unsupervised\nasymmetric distance metric based on cross-view clustering. The asymmetric\ndistance metric allows specific feature transformations for each camera view to\ntackle the specific feature distortions. We then design a novel unsupervised\nloss function to embed the asymmetric metric into a deep neural network, and\ntherefore develop a novel unsupervised deep framework named the DEep\nClustering-based Asymmetric MEtric Learning (DECAMEL). In such a way, DECAMEL\njointly learns the feature representation and the unsupervised asymmetric\nmetric. DECAMEL learns a compact cross-view cluster structure of Re-ID data,\nand thus help alleviate the view-specific bias and facilitate mining the\npotential cross-view discriminative information for unsupervised Re-ID.\nExtensive experiments on seven benchmark datasets whose sizes span several\norders show the effectiveness of our framework.","url_abs":"http://arxiv.org/abs/1901.10177v1","url_pdf":"http://arxiv.org/pdf/1901.10177v1.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":"unsupervised-person-re-identification-by-deep","repo_url":"https://github.com/KovenYu/DECAMEL","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"clustering","task_name":"Clustering"},{"task_slug":"deep-clustering","task_name":"Deep Clustering"},{"task_slug":"metric-learning","task_name":"Metric Learning"},{"task_slug":"person-re-identification","task_name":"Person Re-Identification"},{"task_slug":"unsupervised-person-re-identification","task_name":"Unsupervised Person Re-Identification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1901.10177","atlas_url":"https://app.syntology.ai/?focus=1901.10177","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}