NASA: Neural Articulated Shape Approximation

TL;DR

NASA introduces a neural framework for representing articulated deformable objects that simplifies occupancy testing and enhances efficiency in 3D tracking, offering an alternative to traditional mesh-based methods.

Contribution

The paper presents NASA, a neural indicator function-based approach for efficient articulated shape approximation conditioned on pose, simplifying deformation representation.

Findings

Effective for 3D tracking applications

Circumvents mesh complexity and water-tightness issues

Demonstrates potential for extensions

Abstract

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.