{"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/mvf-net-multi-view-3d-face-morphable-model","title":"MVF-Net: Multi-View 3D Face Morphable Model Regression","arxiv_id":"1904.04473","date":"2019-04-09","proceeding":"CVPR 2019 6","authors":["Fanzi Wu","Linchao Bao","Yajing Chen","Yonggen Ling","Yibing Song","Songnan Li","King Ngi Ngan","Wei Liu"],"abstract":"We address the problem of recovering the 3D geometry of a human face from a\nset of facial images in multiple views. While recent studies have shown\nimpressive progress in 3D Morphable Model (3DMM) based facial reconstruction,\nthe settings are mostly restricted to a single view. There is an inherent\ndrawback in the single-view setting: the lack of reliable 3D constraints can\ncause unresolvable ambiguities. We in this paper explore 3DMM-based shape\nrecovery in a different setting, where a set of multi-view facial images are\ngiven as input. A novel approach is proposed to regress 3DMM parameters from\nmulti-view inputs with an end-to-end trainable Convolutional Neural Network\n(CNN). Multiview geometric constraints are incorporated into the network by\nestablishing dense correspondences between different views leveraging a novel\nself-supervised view alignment loss. The main ingredient of the view alignment\nloss is a differentiable dense optical flow estimator that can backpropagate\nthe alignment errors between an input view and a synthetic rendering from\nanother input view, which is projected to the target view through the 3D shape\nto be inferred. Through minimizing the view alignment loss, better 3D shapes\ncan be recovered such that the synthetic projections from one view to another\ncan better align with the observed image. Extensive experiments demonstrate the\nsuperiority of the proposed method over other 3DMM methods.","url_abs":"http://arxiv.org/abs/1904.04473v1","url_pdf":"http://arxiv.org/pdf/1904.04473v1.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":"mvf-net-multi-view-3d-face-morphable-model","repo_url":"https://github.com/Fanziapril/mvfnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"3d-geometry","task_name":"3D geometry"},{"task_slug":"optical-flow-estimation","task_name":"Optical Flow Estimation"},{"task_slug":"regression-1","task_name":"regression"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1904.04473","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}