{"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/regressing-robust-and-discriminative-3d","title":"Regressing Robust and Discriminative 3D Morphable Models with a very Deep Neural Network","arxiv_id":"1612.04904","date":"2016-12-15","proceeding":"CVPR 2017 7","authors":["Anh Tuan Tran","Tal Hassner","Iacopo Masi","Gerard Medioni"],"abstract":"The 3D shapes of faces are well known to be discriminative. Yet despite this,\nthey are rarely used for face recognition and always under controlled viewing\nconditions. We claim that this is a symptom of a serious but often overlooked\nproblem with existing methods for single view 3D face reconstruction: when\napplied \"in the wild\", their 3D estimates are either unstable and change for\ndifferent photos of the same subject or they are over-regularized and generic.\nIn response, we describe a robust method for regressing discriminative 3D\nmorphable face models (3DMM). We use a convolutional neural network (CNN) to\nregress 3DMM shape and texture parameters directly from an input photo. We\novercome the shortage of training data required for this purpose by offering a\nmethod for generating huge numbers of labeled examples. The 3D estimates\nproduced by our CNN surpass state of the art accuracy on the MICC data set.\nCoupled with a 3D-3D face matching pipeline, we show the first competitive face\nrecognition results on the LFW, YTF and IJB-A benchmarks using 3D face shapes\nas representations, rather than the opaque deep feature vectors used by other\nmodern systems.","url_abs":"http://arxiv.org/abs/1612.04904v1","url_pdf":"http://arxiv.org/pdf/1612.04904v1.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":"regressing-robust-and-discriminative-3d","repo_url":"https://github.com/anhttran/3dmm_basic","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null},{"paper_slug":"regressing-robust-and-discriminative-3d","repo_url":"https://github.com/anhttran/3dmm_cnn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null},{"paper_slug":"regressing-robust-and-discriminative-3d","repo_url":"https://github.com/fengju514/Expression-Net","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"regressing-robust-and-discriminative-3d","repo_url":"https://github.com/sagpant/3dmm_cnn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null},{"paper_slug":"regressing-robust-and-discriminative-3d","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":"3d-face-reconstruction","task_name":"3D Face Reconstruction"},{"task_slug":"face-recognition","task_name":"Face Recognition"},{"task_slug":"face-reconstruction","task_name":"Face Reconstruction"},{"task_slug":"face-verification","task_name":"Face Verification"}],"methods":[{"method_slug":"3d-convolution","method_name":"3D Convolution"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-face-reconstruction-on-florence","task":"3D Face Reconstruction","dataset":"Florence","model":"3DMM-CNN","rank_in_archive_order":7,"of":16,"metrics":{"Average 3D Error":"1.93"},"uses_additional_data":false},{"leaderboard":"/sota/3d-face-reconstruction-on-florence","task":"3D Face Reconstruction","dataset":"Florence","model":"Tran et al.","rank_in_archive_order":15,"of":16,"metrics":{"RMSE Cooperative":"1.97","RMSE Indoor":"2.03","RMSE Outdoor":"1.93"},"uses_additional_data":false},{"leaderboard":"/sota/3d-face-reconstruction-on-now-benchmark-1","task":"3D Face Reconstruction","dataset":"NoW Benchmark","model":"3DMM-CNN","rank_in_archive_order":17,"of":17,"metrics":{"Mean Reconstruction Error (mm)":"2.33","Median Reconstruction Error":"1.84","Stdev Reconstruction Error (mm)":"2.05"},"uses_additional_data":false},{"leaderboard":"/sota/face-verification-on-youtube-faces-db","task":"Face Verification","dataset":"YouTube Faces DB","model":"3DMM face shape parameters + CNN","rank_in_archive_order":12,"of":12,"metrics":{"Accuracy":"88.80%"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1612.04904","atlas_url":"https://app.syntology.ai/?focus=1612.04904","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}