{"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/unifying-identification-and-context-learning","title":"Unifying Identification and Context Learning for Person Recognition","arxiv_id":"1806.03084","date":"2018-06-08","proceeding":"CVPR 2018 6","authors":["Qingqiu Huang","Yu Xiong","Dahua Lin"],"abstract":"Despite the great success of face recognition techniques, recognizing persons\nunder unconstrained settings remains challenging. Issues like profile views,\nunfavorable lighting, and occlusions can cause substantial difficulties.\nPrevious works have attempted to tackle this problem by exploiting the context,\ne.g. clothes and social relations. While showing promising improvement, they\nare usually limited in two important aspects, relying on simple heuristics to\ncombine different cues and separating the construction of context from people\nidentities. In this work, we aim to move beyond such limitations and propose a\nnew framework to leverage context for person recognition. In particular, we\npropose a Region Attention Network, which is learned to adaptively combine\nvisual cues with instance-dependent weights. We also develop a unified\nformulation, where the social contexts are learned along with the reasoning of\npeople identities. These models substantially improve the robustness when\nworking with the complex contextual relations in unconstrained environments. On\ntwo large datasets, PIPA and Cast In Movies (CIM), a new dataset proposed in\nthis work, our method consistently achieves state-of-the-art performance under\nmultiple evaluation policies.","url_abs":"http://arxiv.org/abs/1806.03084v1","url_pdf":"http://arxiv.org/pdf/1806.03084v1.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":"unifying-identification-and-context-learning","repo_url":"https://github.com/ycxioooong/MovieSynopsisAssociation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"face-recognition","task_name":"Face Recognition"},{"task_slug":"person-recognition","task_name":"Person Recognition"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1806.03084","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}