{"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/robust-face-detection-via-learning-small","title":"Robust Face Detection via Learning Small Faces on Hard Images","arxiv_id":"1811.11662","date":"2018-11-28","proceeding":null,"authors":["Zhishuai Zhang","Wei Shen","Siyuan Qiao","Yan Wang","Bo wang","Alan Yuille"],"abstract":"Recent anchor-based deep face detectors have achieved promising performance,\nbut they are still struggling to detect hard faces, such as small, blurred and\npartially occluded faces. A reason is that they treat all images and faces\nequally, without putting more effort on hard ones; however, many training\nimages only contain easy faces, which are less helpful to achieve better\nperformance on hard images. In this paper, we propose that the robustness of a\nface detector against hard faces can be improved by learning small faces on\nhard images. Our intuitions are (1) hard images are the images which contain at\nleast one hard face, thus they facilitate training robust face detectors; (2)\nmost hard faces are small faces and other types of hard faces can be easily\nconverted to small faces by shrinking. We build an anchor-based deep face\ndetector, which only output a single feature map with small anchors, to\nspecifically learn small faces and train it by a novel hard image mining\nstrategy. Extensive experiments have been conducted on WIDER FACE, FDDB, Pascal\nFaces, and AFW datasets to show the effectiveness of our method. Our method\nachieves APs of 95.7, 94.9 and 89.7 on easy, medium and hard WIDER FACE val\ndataset respectively, which surpass the previous state-of-the-arts, especially\non the hard subset. Code and model are available at\nhttps://github.com/bairdzhang/smallhardface.","url_abs":"http://arxiv.org/abs/1811.11662v1","url_pdf":"http://arxiv.org/pdf/1811.11662v1.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":"robust-face-detection-via-learning-small","repo_url":"https://github.com/bairdzhang/smallhardface","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"face-detection","task_name":"Face Detection"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/face-detection-on-wider-face-easy","task":"Face Detection","dataset":"WIDER Face (Easy)","model":"+ DH + HIM","rank_in_archive_order":11,"of":27,"metrics":{"AP":"0.957"},"uses_additional_data":false},{"leaderboard":"/sota/face-detection-on-wider-face-hard","task":"Face Detection","dataset":"WIDER Face (Hard)","model":"+ DH + HIM","rank_in_archive_order":7,"of":40,"metrics":{"AP":"0.897"},"uses_additional_data":false},{"leaderboard":"/sota/face-detection-on-wider-face-medium","task":"Face Detection","dataset":"WIDER Face (Medium)","model":"+ DH + HIM","rank_in_archive_order":8,"of":37,"metrics":{"AP":"0.949"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}