NRDF: Neural Riemannian Distance Fields for learning articulated pose priors
Author(s)
He, Yannan
Tiwari, Garvita
Birdal, Tolga
Lenssen, Jan Eric
Pons-Moll, Gerard
Type
Conference Paper
Abstract
Faithfully modeling the space of articulations is a cru-cial task that allows recovery and generation of realistic poses, and remains a notorious challenge. To this end, we introduce Neural Riemannian Distance Fields (NRDFs), data-driven priors modeling the space of plausible articu-lations, represented as the zero-level-set of a neural field in a high-dimensional product-quaternion space. To train NRDFs only on positive examples, we introduce a new sam-pling algorithm, ensuring that the geodesic distances fol-low a desired distribution, yielding a principled distance field learning paradigm. We then devise a projection al-gorithm to map any random pose onto the level-set by an adaptive-step Riemannian optimizer, adhering to the product manifold of joint rotations at all times. NRDFs can compute the Riemannian gradient via backpropagation and by mathematical analogy, are related to Rieman-nian flow matching, a recent generative model. We conduct a comprehensive evaluation of NRDF against other pose priors in various downstream tasks, i.e., pose generation, image-based pose estimation, and solving inverse kinematics, highlighting NRDF's superior performance. Besides humans, NRDF's versatility extends to hand and animal poses, as it can effectively represent any articulation.
Date Issued
2024-06-16
Date Acceptance
2024-06-01
Citation
2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024, pp.1661-1671
Publisher
IEEE
Start Page
1661
End Page
1671
Journal / Book Title
2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
Copyright Statement
©2024 The Author(s). This CVPR paper is the Open Access version, provided by the Computer Vision Foundation.
Except for this watermark, it is identical to the accepted version; the final published version of the proceedings is available on IEEE Xplore.
Except for this watermark, it is identical to the accepted version; the final published version of the proceedings is available on IEEE Xplore.
Identifier
https://doi.org/10.1109/cvpr52733.2024.00164
Source
Conference on Computer Vision and Pattern Recognition (CVPR)
Publication Status
Published
Start Date
2024-06-16
Finish Date
2024-06-22
Coverage Spatial
Seattle, WA, USA