BaRe-ESA: A Riemannian Framework for Unregistered Human Body Shapes

TL;DR

BaRe-ESA introduces a Riemannian framework for unregistered human body shapes that improves shape registration, interpolation, and extrapolation without requiring point correspondences, demonstrated on FAUST datasets.

Contribution

It presents a novel Riemannian approach operating directly on unregistered meshes, enabling better shape analysis without prior correspondences.

Findings

Significant improvements in shape registration, interpolation, and extrapolation.

Effective in motion transfer and random body shape generation.

Operates efficiently on unregistered mesh data.

Abstract

We present Basis Restricted Elastic Shape Analysis (BaRe-ESA), a novel Riemannian framework for human body scan representation, interpolation and extrapolation. BaRe-ESA operates directly on unregistered meshes, i.e., without the need to establish prior point to point correspondences or to assume a consistent mesh structure. Our method relies on a latent space representation, which is equipped with a Riemannian (non-Euclidean) metric associated to an invariant higher-order metric on the space of surfaces. Experimental results on the FAUST and DFAUST datasets show that BaRe-ESA brings significant improvements with respect to previous solutions in terms of shape registration, interpolation and extrapolation. The efficiency and strength of our model is further demonstrated in applications such as motion transfer and random generation of body shape and pose.