{"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/face-alignment-across-large-poses-a-3d","title":"Face Alignment Across Large Poses: A 3D Solution","arxiv_id":"1511.07212","date":"2015-11-23","proceeding":"CVPR 2016 6","authors":["Xiangyu Zhu","Zhen Lei","Xiaoming Liu","Hailin Shi","Stan Z. Li"],"abstract":"Face alignment, which fits a face model to an image and extracts the semantic\nmeanings of facial pixels, has been an important topic in CV community.\nHowever, most algorithms are designed for faces in small to medium poses (below\n45 degree), lacking the ability to align faces in large poses up to 90 degree.\nThe challenges are three-fold: Firstly, the commonly used landmark-based face\nmodel assumes that all the landmarks are visible and is therefore not suitable\nfor profile views. Secondly, the face appearance varies more dramatically\nacross large poses, ranging from frontal view to profile view. Thirdly,\nlabelling landmarks in large poses is extremely challenging since the invisible\nlandmarks have to be guessed. In this paper, we propose a solution to the three\nproblems in an new alignment framework, called 3D Dense Face Alignment (3DDFA),\nin which a dense 3D face model is fitted to the image via convolutional neutral\nnetwork (CNN). We also propose a method to synthesize large-scale training\nsamples in profile views to solve the third problem of data labelling.\nExperiments on the challenging AFLW database show that our approach achieves\nsignificant improvements over state-of-the-art methods.","url_abs":"http://arxiv.org/abs/1511.07212v1","url_pdf":"http://arxiv.org/pdf/1511.07212v1.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":"3d-face-reconstruction","task_name":"3D Face Reconstruction"},{"task_slug":"face-alignment","task_name":"Face Alignment"},{"task_slug":"face-model","task_name":"Face Model"},{"task_slug":"head-pose-estimation","task_name":"Head Pose Estimation"}],"methods":[],"datasets_introduced":[{"slug":"aflw2000-3d","name":"AFLW2000-3D","full_name":""}],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-face-reconstruction-on-aflw2000-3d","task":"3D Face Reconstruction","dataset":"AFLW2000-3D","model":"3DDFA","rank_in_archive_order":6,"of":8,"metrics":{"Mean NME ":"5.3695%"},"uses_additional_data":false},{"leaderboard":"/sota/3d-face-reconstruction-on-florence","task":"3D Face Reconstruction","dataset":"Florence","model":"3DDFA","rank_in_archive_order":3,"of":16,"metrics":{"Mean NME ":"6.3833%"},"uses_additional_data":false},{"leaderboard":"/sota/face-alignment-on-aflw2000","task":"Face Alignment","dataset":"AFLW2000","model":"3DDFA","rank_in_archive_order":4,"of":5,"metrics":{"Error rate":"5.42"},"uses_additional_data":false},{"leaderboard":"/sota/face-alignment-on-aflw2000-3d","task":"Face Alignment","dataset":"AFLW2000-3D","model":"3DDFA + SDM","rank_in_archive_order":13,"of":14,"metrics":{"Balanced NME (2D Sparse Alignment)":"4.94%"},"uses_additional_data":false},{"leaderboard":"/sota/facial-landmark-detection-on-300w","task":"Facial Landmark Detection","dataset":"300W","model":"CFSS","rank_in_archive_order":12,"of":15,"metrics":{"NME":"5.76"},"uses_additional_data":false},{"leaderboard":"/sota/facial-landmark-detection-on-300w","task":"Facial Landmark Detection","dataset":"300W","model":"3DDFA","rank_in_archive_order":14,"of":15,"metrics":{"NME":"7.01"},"uses_additional_data":false},{"leaderboard":"/sota/head-pose-estimation-on-aflw2000","task":"Head Pose Estimation","dataset":"AFLW2000","model":"3DDFA","rank_in_archive_order":23,"of":25,"metrics":{"MAE":" 7.393"},"uses_additional_data":false},{"leaderboard":"/sota/head-pose-estimation-on-biwi","task":"Head Pose Estimation","dataset":"BIWI","model":"3DDFA","rank_in_archive_order":21,"of":29,"metrics":{"MAE (trained with other data)":"19.068"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}