{"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/190409290","title":"FeatherNets: Convolutional Neural Networks as Light as Feather for Face Anti-spoofing","arxiv_id":"1904.09290","date":"2019-04-22","proceeding":null,"authors":["Peng Zhang","Fuhao Zou","Zhiwen Wu","Nengli Dai","Skarpness Mark","Michael Fu","Juan Zhao","Kai Li"],"abstract":"Face Anti-spoofing gains increased attentions recently in both academic and\nindustrial fields. With the emergence of various CNN based solutions, the\nmulti-modal(RGB, depth and IR) methods based CNN showed better performance than\nsingle modal classifiers. However, there is a need for improving the\nperformance and reducing the complexity. Therefore, an extreme light network\narchitecture(FeatherNet A/B) is proposed with a streaming module which fixes\nthe weakness of Global Average Pooling and uses less parameters. Our single\nFeatherNet trained by depth image only, provides a higher baseline with 0.00168\nACER, 0.35M parameters and 83M FLOPS. Furthermore, a novel fusion procedure\nwith ``ensemble + cascade'' structure is presented to satisfy the performance\npreferred use cases. Meanwhile, the MMFD dataset is collected to provide more\nattacks and diversity to gain better generalization. We use the fusion method\nin the Face Anti-spoofing Attack Detection Challenge@CVPR2019 and got the\nresult of 0.0013(ACER), 0.999(TPR@FPR=10e-2), 0.998(TPR@FPR=10e-3) and\n0.9814(TPR@FPR=10e-4).","url_abs":"http://arxiv.org/abs/1904.09290v1","url_pdf":"http://arxiv.org/pdf/1904.09290v1.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":"190409290","repo_url":"https://github.com/SoftwareGift/FeatheNets_Face-Anti-spoofing-Attack-Detection-Challenge-CVPR2019","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"190409290","repo_url":"https://github.com/fangzong12/FeatherNet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"190409290","repo_url":"https://github.com/Recognito-Vision/Android-FaceRecognition-FaceLivenessDetection","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"diversity","task_name":"Diversity"},{"task_slug":"face-anti-spoofing","task_name":"Face Anti-Spoofing"}],"methods":[{"method_slug":"streaming-module","method_name":"Streaming Module"}],"datasets_introduced":[],"methods_introduced":[{"slug":"streaming-module","name":"Streaming Module","full_name":"Streaming Module"}],"results":[{"leaderboard":"/sota/face-anti-spoofing-on-celeba-spoof-enroll5","task":"Face Anti-Spoofing","dataset":"CelebA-Spoof-Enroll5","model":"FeatherNet","rank_in_archive_order":5,"of":5,"metrics":{"AUC":"97.1"},"uses_additional_data":false},{"leaderboard":"/sota/face-anti-spoofing-on-siw-protocol-3","task":"Face Anti-Spoofing","dataset":"SiW (Protocol 3)","model":"FeatherNet","rank_in_archive_order":6,"of":7,"metrics":{"ACER":"31.1"},"uses_additional_data":false},{"leaderboard":"/sota/face-anti-spoofing-on-siw-enroll5","task":"Face Anti-Spoofing","dataset":"SiW-Enroll5","model":"FeatherNet","rank_in_archive_order":3,"of":5,"metrics":{"AUC":"98.9"},"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}