{"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/template-adaptation-for-face-verification-and","title":"Template Adaptation for Face Verification and Identification","arxiv_id":"1603.03958","date":"2016-03-12","proceeding":null,"authors":["Nate Crosswhite","Jeffrey Byrne","Omkar M. Parkhi","Chris Stauffer","Qiong Cao","Andrew Zisserman"],"abstract":"Face recognition performance evaluation has traditionally focused on\none-to-one verification, popularized by the Labeled Faces in the Wild dataset\nfor imagery and the YouTubeFaces dataset for videos. In contrast, the newly\nreleased IJB-A face recognition dataset unifies evaluation of one-to-many face\nidentification with one-to-one face verification over templates, or sets of\nimagery and videos for a subject. In this paper, we study the problem of\ntemplate adaptation, a form of transfer learning to the set of media in a\ntemplate. Extensive performance evaluations on IJB-A show a surprising result,\nthat perhaps the simplest method of template adaptation, combining deep\nconvolutional network features with template specific linear SVMs, outperforms\nthe state-of-the-art by a wide margin. We study the effects of template size,\nnegative set construction and classifier fusion on performance, then compare\ntemplate adaptation to convolutional networks with metric learning, 2D and 3D\nalignment. Our unexpected conclusion is that these other methods, when combined\nwith template adaptation, all achieve nearly the same top performance on IJB-A\nfor template-based face verification and identification.","url_abs":"http://arxiv.org/abs/1603.03958v3","url_pdf":"http://arxiv.org/pdf/1603.03958v3.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":"face-identification","task_name":"Face Identification"},{"task_slug":"face-recognition","task_name":"Face Recognition"},{"task_slug":"face-verification","task_name":"Face Verification"},{"task_slug":"metric-learning","task_name":"Metric Learning"},{"task_slug":"transfer-learning","task_name":"Transfer Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/face-verification-on-ijb-a","task":"Face Verification","dataset":"IJB-A","model":"Template adaptation","rank_in_archive_order":9,"of":17,"metrics":{"TAR @ FAR=0.01":"93.90%"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1603.03958","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}