NASA: Neural Articulated Shape Approximation

Efficient representation of articulated objects such as human bodies is an important problem in computer vision and graphics. To efficiently simulate deformation, existing approaches represent 3D objects using polygonal meshes and deform them using skinning techniques. This paper introduces neural a...

Full description

Saved in:
Bibliographic Details
Published inarXiv.org
Main Authors Deng, Boyang, Lewis, J P, Jeruzalski, Timothy, Pons-Moll, Gerard, Hinton, Geoffrey, Norouzi, Mohammad, Tagliasacchi, Andrea
Format Paper
LanguageEnglish
Published Ithaca Cornell University Library, arXiv.org 21.07.2022
Subjects
Approximation
Computer simulation
Computer vision
Deformation
Formability
Mathematical analysis
Occupancy
Octrees
Representations
Watertightness
Online AccessGet full text

Cover

More Information
Summary:Efficient representation of articulated objects such as human bodies is an important problem in computer vision and graphics. To efficiently simulate deformation, existing approaches represent 3D objects using polygonal meshes and deform them using skinning techniques. This paper introduces neural articulated shape approximation (NASA), an alternative framework that enables efficient representation of articulated deformable objects using neural indicator functions that are conditioned on pose. Occupancy testing using NASA is straightforward, circumventing the complexity of meshes and the issue of water-tightness. We demonstrate the effectiveness of NASA for 3D tracking applications, and discuss other potential extensions.
ISSN:2331-8422