Few-Shot High-Fidelity 3D Reconstruction of Human Bodies - Abstract 21.63
J. Garcia, "Few-Shot High-Fidelity 3D Reconstruction of Human Bodies", Proc. of 3DBODY.TECH 2021 - 12th Int. Conf. and Exh. on 3D Body Scanning and Processing Technologies, Lugano, Switzerland, 19-20 Oct. 2021, #63.
Title:
Few-Shot High-Fidelity 3D Reconstruction of Human Bodies
Authors:
Jaime GARCIA
Crisalix SA, Lausanne, Switzerland
Abstract:
We provide a fully automated solution for the 3D reconstruction of human heads, including hair and shoulders, from a sparse collection of photos (as few as 3). Our approach yields anatomically accurate reconstructions, improving over commonly used methods based on 3D morphable models, while retaining a comparable robustness. The key technology enabling our method is a deep learning system that combines a 3D prior (shape space) pretrained from over 10K real scans, with a state-of-the art 3D reconstruction architecture based on differentiable rendering. This technology offers a promising avenue to model and reconstruct full bodies in 3D with unprecedented accuracy and flexibility, being able to capture complex shapes, e.g garments, clothing, or diverse bodies.
Abstract:
Presentation:
Details:
Proceedings: 3DBODY.TECH 2021, 19-20 Oct. 2021, Lugano, Switzerland
Paper id#: 63
License/Copyright notice
Proceedings: © Hometrica Consulting - Dr. Nicola D'Apuzzo, Switzerland, hometrica.ch.
Authors retain all rights to individual papers, which are licensed under a Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
The papers appearing in the proceedings reflect the author's opinions. Their inclusion in the proceedings does not necessary constitute endorsement by the editor or by the publisher.
Note: click the + on the top left of the page to open/close the menu.