{"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/large-pose-3d-face-reconstruction-from-a","title":"Large Pose 3D Face Reconstruction from a Single Image via Direct Volumetric CNN Regression","arxiv_id":"1703.07834","date":"2017-03-22","proceeding":"ICCV 2017 10","authors":["Aaron S. Jackson","Adrian Bulat","Vasileios Argyriou","Georgios Tzimiropoulos"],"abstract":"3D face reconstruction is a fundamental Computer Vision problem of\nextraordinary difficulty. Current systems often assume the availability of\nmultiple facial images (sometimes from the same subject) as input, and must\naddress a number of methodological challenges such as establishing dense\ncorrespondences across large facial poses, expressions, and non-uniform\nillumination. In general these methods require complex and inefficient\npipelines for model building and fitting. In this work, we propose to address\nmany of these limitations by training a Convolutional Neural Network (CNN) on\nan appropriate dataset consisting of 2D images and 3D facial models or scans.\nOur CNN works with just a single 2D facial image, does not require accurate\nalignment nor establishes dense correspondence between images, works for\narbitrary facial poses and expressions, and can be used to reconstruct the\nwhole 3D facial geometry (including the non-visible parts of the face)\nbypassing the construction (during training) and fitting (during testing) of a\n3D Morphable Model. We achieve this via a simple CNN architecture that performs\ndirect regression of a volumetric representation of the 3D facial geometry from\na single 2D image. We also demonstrate how the related task of facial landmark\nlocalization can be incorporated into the proposed framework and help improve\nreconstruction quality, especially for the cases of large poses and facial\nexpressions. Testing code will be made available online, along with pre-trained\nmodels http://aaronsplace.co.uk/papers/jackson2017recon","url_abs":"http://arxiv.org/abs/1703.07834v2","url_pdf":"http://arxiv.org/pdf/1703.07834v2.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":"large-pose-3d-face-reconstruction-from-a","repo_url":"https://github.com/AaronJackson/vrn","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"torch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"3d-face-reconstruction","task_name":"3D Face Reconstruction"},{"task_slug":"face-alignment","task_name":"Face Alignment"},{"task_slug":"face-reconstruction","task_name":"Face Reconstruction"},{"task_slug":"regression-1","task_name":"regression"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-face-reconstruction-on-florence","task":"3D Face Reconstruction","dataset":"Florence","model":"VRN-Guided","rank_in_archive_order":2,"of":16,"metrics":{"Mean NME ":"5.2667%"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1703.07834","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}