{"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-pre-training-for-person-re","title":"Unsupervised Pre-training for Person Re-identification","arxiv_id":"2012.03753","date":"2020-12-07","proceeding":"CVPR 2021 1","authors":["Dengpan Fu","Dongdong Chen","Jianmin Bao","Hao Yang","Lu Yuan","Lei Zhang","Houqiang Li","Dong Chen"],"abstract":"In this paper, we present a large scale unlabeled person re-identification (Re-ID) dataset \"LUPerson\" and make the first attempt of performing unsupervised pre-training for improving the generalization ability of the learned person Re-ID feature representation. This is to address the problem that all existing person Re-ID datasets are all of limited scale due to the costly effort required for data annotation. Previous research tries to leverage models pre-trained on ImageNet to mitigate the shortage of person Re-ID data but suffers from the large domain gap between ImageNet and person Re-ID data. LUPerson is an unlabeled dataset of 4M images of over 200K identities, which is 30X larger than the largest existing Re-ID dataset. It also covers a much diverse range of capturing environments (eg, camera settings, scenes, etc.). Based on this dataset, we systematically study the key factors for learning Re-ID features from two perspectives: data augmentation and contrastive loss. Unsupervised pre-training performed on this large-scale dataset effectively leads to a generic Re-ID feature that can benefit all existing person Re-ID methods. Using our pre-trained model in some basic frameworks, our methods achieve state-of-the-art results without bells and whistles on four widely used Re-ID datasets: CUHK03, Market1501, DukeMTMC, and MSMT17. Our results also show that the performance improvement is more significant on small-scale target datasets or under few-shot setting.","url_abs":"https://arxiv.org/abs/2012.03753v2","url_pdf":"https://arxiv.org/pdf/2012.03753v2.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-pre-training-for-person-re","repo_url":"https://github.com/DengpanFu/LUPerson","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"data-augmentation","task_name":"Data Augmentation"},{"task_slug":"person-re-identification","task_name":"Person Re-Identification"},{"task_slug":"unsupervised-pre-training","task_name":"Unsupervised Pre-training"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/person-re-identification-on-cuhk03","task":"Person Re-Identification","dataset":"CUHK03","model":"Unsupervised Pre-training (ResNet50+BDB)","rank_in_archive_order":9,"of":19,"metrics":{"MAP":"79.6","Rank-1":"81.9"},"uses_additional_data":false},{"leaderboard":"/sota/person-re-identification-on-dukemtmc-reid","task":"Person Re-Identification","dataset":"DukeMTMC-reID","model":"Unsupervised Pre-training (ResNet101+RK)","rank_in_archive_order":4,"of":94,"metrics":{"Rank-1":"93.99","mAP":"92.77"},"uses_additional_data":true},{"leaderboard":"/sota/person-re-identification-on-dukemtmc-reid","task":"Person Re-Identification","dataset":"DukeMTMC-reID","model":"Unsupervised Pre-training (ResNet101+MGN)","rank_in_archive_order":25,"of":94,"metrics":{"Rank-1":"91.9","mAP":"84.1"},"uses_additional_data":false},{"leaderboard":"/sota/person-re-identification-on-msmt17","task":"Person Re-Identification","dataset":"MSMT17","model":"Unsupervised Pre-training (ResNet101+MGN)","rank_in_archive_order":15,"of":43,"metrics":{"Rank-1":"86.6","mAP":"68.8"},"uses_additional_data":true},{"leaderboard":"/sota/person-re-identification-on-market-1501","task":"Person Re-Identification","dataset":"Market-1501","model":"Unsupervised Pre-training (ResNet101+MGN)","rank_in_archive_order":5,"of":135,"metrics":{"Rank-1":"97","mAP":"92"},"uses_additional_data":false},{"leaderboard":"/sota/person-re-identification-on-market-1501","task":"Person Re-Identification","dataset":"Market-1501","model":"Unsupervised Pre-training (ResNet101+RK)","rank_in_archive_order":126,"of":135,"metrics":{"mAP":"96.21"},"uses_additional_data":false},{"leaderboard":"/sota/person-re-identification-on-market-1501-c","task":"Person Re-Identification","dataset":"Market-1501-C","model":"LUPerson","rank_in_archive_order":22,"of":22,"metrics":{"Rank-1":"32.22","mAP":"10.37","mINP":"0.29"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/2012.03753","atlas_url":"https://app.syntology.ai/?focus=2012.03753","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}