{"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/pfld-a-practical-facial-landmark-detector","title":"PFLD: A Practical Facial Landmark Detector","arxiv_id":"1902.10859","date":"2019-02-28","proceeding":null,"authors":["Xiaojie Guo","Siyuan Li","Jinke Yu","Jiawan Zhang","Jiayi Ma","Lin Ma","Wei Liu","Haibin Ling"],"abstract":"Being accurate, efficient, and compact is essential to a facial landmark\ndetector for practical use. To simultaneously consider the three concerns, this\npaper investigates a neat model with promising detection accuracy under wild\nenvironments e.g., unconstrained pose, expression, lighting, and occlusion\nconditions) and super real-time speed on a mobile device. More concretely, we\ncustomize an end-to-end single stage network associated with acceleration\ntechniques. During the training phase, for each sample, rotation information is\nestimated for geometrically regularizing landmark localization, which is then\nNOT involved in the testing phase. A novel loss is designed to, besides\nconsidering the geometrical regularization, mitigate the issue of data\nimbalance by adjusting weights of samples to different states, such as large\npose, extreme lighting, and occlusion, in the training set. Extensive\nexperiments are conducted to demonstrate the efficacy of our design and reveal\nits superior performance over state-of-the-art alternatives on widely-adopted\nchallenging benchmarks, i.e., 300W (including iBUG, LFPW, AFW, HELEN, and\nXM2VTS) and AFLW. Our model can be merely 2.1Mb of size and reach over 140 fps\nper face on a mobile phone (Qualcomm ARM 845 processor) with high precision,\nmaking it attractive for large-scale or real-time applications. We have made\nour practical system based on PFLD 0.25X model publicly available at\n\\url{http://sites.google.com/view/xjguo/fld} for encouraging comparisons and\nimprovements from the community.","url_abs":"http://arxiv.org/abs/1902.10859v2","url_pdf":"http://arxiv.org/pdf/1902.10859v2.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":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/AmrElsersy/PFLD-Pytorch-Landmarks","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/Ontheway361/pfld-pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/RomanticWithoutStatus/PFLD-TensorFlow2caffe","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/SpikeKing/Gaze4Landmarks","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/ainrichman/Peppa-Facial-Landmark-PyTorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/epoc88/PFLD_68pts_Pytorch_2020","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"mxnet","reach":{"status":"ok"}},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/github-luffy/PFLD-68points-tensorflow","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"mxnet","reach":null},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/github-luffy/PFLD_68points_Pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"mxnet","reach":null},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/guoqiangqi/PFLD","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/nilseuropa/pfdl_ncnn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/nilseuropa/pfld_ncnn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/olegpolivin/antisleep","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/polarisZhao/PFLD-pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/samuelyu2002/PFLD","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/starhiking/ATF","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/SongJieLiu/pfld","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/xiuyu0000/papers_with_examples/tree/main/pfld","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null},{"paper_slug":"pfld-a-practical-facial-landmark-detector","repo_url":"https://github.com/yangyucheng000/pfld","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null}],"tasks":[{"task_slug":"face-alignment","task_name":"Face Alignment"},{"task_slug":"facial-landmark-detection","task_name":"Facial Landmark Detection"}],"methods":[{"method_slug":"speed","method_name":"SPEED"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1902.10859","atlas_url":"https://app.syntology.ai/?focus=1902.10859","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}