{"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/recognizing-disguised-faces-in-the-wild","title":"Recognizing Disguised Faces in the Wild","arxiv_id":"1811.08837","date":"2018-11-21","proceeding":null,"authors":["Maneet Singh","Richa Singh","Mayank Vatsa","Nalini Ratha","Rama Chellappa"],"abstract":"Research in face recognition has seen tremendous growth over the past couple\nof decades. Beginning from algorithms capable of performing recognition in\nconstrained environments, the current face recognition systems achieve very\nhigh accuracies on large-scale unconstrained face datasets. While upcoming\nalgorithms continue to achieve improved performance, a majority of the face\nrecognition systems are susceptible to failure under disguise variations, one\nof the most challenging covariate of face recognition. Most of the existing\ndisguise datasets contain images with limited variations, often captured in\ncontrolled settings. This does not simulate a real world scenario, where both\nintentional and unintentional unconstrained disguises are encountered by a face\nrecognition system. In this paper, a novel Disguised Faces in the Wild (DFW)\ndataset is proposed which contains over 11000 images of 1000 identities with\ndifferent types of disguise accessories. The dataset is collected from the\nInternet, resulting in unconstrained face images similar to real world\nsettings. This is the first-of-a-kind dataset with the availability of\nimpersonator and genuine obfuscated face images for each subject. The proposed\ndataset has been analyzed in terms of three levels of difficulty: (i) easy,\n(ii) medium, and (iii) hard in order to showcase the challenging nature of the\nproblem. It is our view that the research community can greatly benefit from\nthe DFW dataset in terms of developing algorithms robust to such adversaries.\nThe proposed dataset was released as part of the First International Workshop\nand Competition on Disguised Faces in the Wild at CVPR, 2018. This paper\npresents the DFW dataset in detail, including the evaluation protocols,\nbaseline results, performance analysis of the submissions received as part of\nthe competition, and three levels of difficulties of the DFW challenge dataset.","url_abs":"http://arxiv.org/abs/1811.08837v1","url_pdf":"http://arxiv.org/pdf/1811.08837v1.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":"disguised-face-verification","task_name":"Disguised Face Verification"},{"task_slug":"face-recognition","task_name":"Face Recognition"}],"methods":[],"datasets_introduced":[{"slug":"dfw","name":"DFW","full_name":"Disguised Faces in the Wild"}],"methods_introduced":[],"results":[{"leaderboard":"/sota/disguised-face-verification-on-disguised","task":"Disguised Face Verification","dataset":"Disguised Faces in the Wild","model":"VGG-Face model features + cosine similarity metric","rank_in_archive_order":2,"of":2,"metrics":{"GAR @0.1% FAR":"17.73","GAR @1% FAR":"33.76"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=1811.08837","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}