Large pose 3D face reconstruction from a single image via direct volumetric CNN regression

3D face reconstruction is a fundamental Computer Vision problem of extraordinary difficulty. Current systems often assume the availability of multiple facial images (sometimes from the same subject) as input, and must address a number of methodological challenges such as establishing dense correspon...

Full description

Bibliographic Details
Main Authors: Jackson, Aaron S., Bulat, Adrian, Argyriou, Vasileios, Tzimiropoulos, Georgios
Format: Conference or Workshop Item
Published: 2017
Online Access:https://eprints.nottingham.ac.uk/44754/
_version_ 1848796990764220416
author Jackson, Aaron S.
Bulat, Adrian
Argyriou, Vasileios
Tzimiropoulos, Georgios
author_facet Jackson, Aaron S.
Bulat, Adrian
Argyriou, Vasileios
Tzimiropoulos, Georgios
author_sort Jackson, Aaron S.
building Nottingham Research Data Repository
collection Online Access
description 3D face reconstruction is a fundamental Computer Vision problem of extraordinary difficulty. Current systems often assume the availability of multiple facial images (sometimes from the same subject) as input, and must address a number of methodological challenges such as establishing dense correspondences across large facial poses, expressions, and non-uniform illumination. In general these methods require complex and inefficient pipelines for model building and fitting. In this work, we propose to address many of these limitations by training a Convolutional Neural Network (CNN) on an appropriate dataset consisting of 2D images and 3D facial models or scans. Our CNN works with just a single 2D facial image, does not require accurate alignment nor establishes dense correspondence between images, works for arbitrary facial poses and expressions, and can be used to reconstruct the whole 3D facial geometry (including the non-visible parts of the face) bypassing the construction (during training) and fitting (during testing) of a 3D Morphable Model. We achieve this via a simple CNN architecture that performs direct regression of a volumetric representation of the 3D facial geometry from a single 2D image. We also demonstrate how the related task of facial landmark localization can be incorporated into the proposed framework and help improve reconstruction quality, especially for the cases of large poses and facial expressions. Code and models will be made available at http://aaronsplace.co.uk
first_indexed 2025-11-14T19:56:46Z
format Conference or Workshop Item
id nottingham-44754
institution University of Nottingham Malaysia Campus
institution_category Local University
last_indexed 2025-11-14T19:56:46Z
publishDate 2017
recordtype eprints
repository_type Digital Repository
spelling nottingham-447542020-05-04T19:14:01Z https://eprints.nottingham.ac.uk/44754/ Large pose 3D face reconstruction from a single image via direct volumetric CNN regression Jackson, Aaron S. Bulat, Adrian Argyriou, Vasileios Tzimiropoulos, Georgios 3D face reconstruction is a fundamental Computer Vision problem of extraordinary difficulty. Current systems often assume the availability of multiple facial images (sometimes from the same subject) as input, and must address a number of methodological challenges such as establishing dense correspondences across large facial poses, expressions, and non-uniform illumination. In general these methods require complex and inefficient pipelines for model building and fitting. In this work, we propose to address many of these limitations by training a Convolutional Neural Network (CNN) on an appropriate dataset consisting of 2D images and 3D facial models or scans. Our CNN works with just a single 2D facial image, does not require accurate alignment nor establishes dense correspondence between images, works for arbitrary facial poses and expressions, and can be used to reconstruct the whole 3D facial geometry (including the non-visible parts of the face) bypassing the construction (during training) and fitting (during testing) of a 3D Morphable Model. We achieve this via a simple CNN architecture that performs direct regression of a volumetric representation of the 3D facial geometry from a single 2D image. We also demonstrate how the related task of facial landmark localization can be incorporated into the proposed framework and help improve reconstruction quality, especially for the cases of large poses and facial expressions. Code and models will be made available at http://aaronsplace.co.uk 2017-10-24 Conference or Workshop Item PeerReviewed Jackson, Aaron S., Bulat, Adrian, Argyriou, Vasileios and Tzimiropoulos, Georgios (2017) Large pose 3D face reconstruction from a single image via direct volumetric CNN regression. In: International Conference on Computer Vision (ICCV17), 22-29 Oct 2017, Venice, Italy. http://ieeexplore.ieee.org/document/8237379/
spellingShingle Jackson, Aaron S.
Bulat, Adrian
Argyriou, Vasileios
Tzimiropoulos, Georgios
Large pose 3D face reconstruction from a single image via direct volumetric CNN regression
title Large pose 3D face reconstruction from a single image via direct volumetric CNN regression
title_full Large pose 3D face reconstruction from a single image via direct volumetric CNN regression
title_fullStr Large pose 3D face reconstruction from a single image via direct volumetric CNN regression
title_full_unstemmed Large pose 3D face reconstruction from a single image via direct volumetric CNN regression
title_short Large pose 3D face reconstruction from a single image via direct volumetric CNN regression
title_sort large pose 3d face reconstruction from a single image via direct volumetric cnn regression
url https://eprints.nottingham.ac.uk/44754/
https://eprints.nottingham.ac.uk/44754/