{"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/domain-adaptation-through-synthesis-for","title":"Domain Adaptation through Synthesis for Unsupervised Person Re-identification","arxiv_id":"1804.10094","date":"2018-04-26","proceeding":"ECCV 2018 9","authors":["Slawomir Bak","Peter Carr","Jean-Francois Lalonde"],"abstract":"Drastic variations in illumination across surveillance cameras make the\nperson re-identification problem extremely challenging. Current large scale\nre-identification datasets have a significant number of training subjects, but\nlack diversity in lighting conditions. As a result, a trained model requires\nfine-tuning to become effective under an unseen illumination condition. To\nalleviate this problem, we introduce a new synthetic dataset that contains\nhundreds of illumination conditions. Specifically, we use 100 virtual humans\nilluminated with multiple HDR environment maps which accurately model realistic\nindoor and outdoor lighting. To achieve better accuracy in unseen illumination\nconditions we propose a novel domain adaptation technique that takes advantage\nof our synthetic data and performs fine-tuning in a completely unsupervised\nway. Our approach yields significantly higher accuracy than semi-supervised and\nunsupervised state-of-the-art methods, and is very competitive with supervised\ntechniques.","url_abs":"http://arxiv.org/abs/1804.10094v1","url_pdf":"http://arxiv.org/pdf/1804.10094v1.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":[],"tasks":[{"task_slug":"diversity","task_name":"Diversity"},{"task_slug":"domain-adaptation","task_name":"Domain Adaptation"},{"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":[{"leaderboard":"/sota/person-re-identification-on-prid2011","task":"Person Re-Identification","dataset":"PRID2011","model":"DASy*","rank_in_archive_order":13,"of":13,"metrics":{"Rank-1":"43.0"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1804.10094","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}