Canonical Pose Reconstruction from Single Depth Image for 3D Non-rigid Pose Recovery on Limited Datasets

TL;DR

This paper introduces a novel canonical pose reconstruction model that converts single depth images of deformable objects into a standard form, enabling effective 3D shape and pose recovery with limited data, outperforming existing methods.

Contribution

The study presents a new approach for non-rigid 3D pose reconstruction from single depth images that works efficiently with small datasets, unlike traditional methods requiring extensive data.

Findings

Effective reconstruction with only ~300 samples

Outperforms state-of-the-art methods on animal and human datasets

Supports pose recovery in voxel representation

Abstract

3D reconstruction from 2D inputs, especially for non-rigid objects like humans, presents unique challenges due to the significant range of possible deformations. Traditional methods often struggle with non-rigid shapes, which require extensive training data to cover the entire deformation space. This study addresses these limitations by proposing a canonical pose reconstruction model that transforms single-view depth images of deformable shapes into a canonical form. This alignment facilitates shape reconstruction by enabling the application of rigid object reconstruction techniques, and supports recovering the input pose in voxel representation as part of the reconstruction task, utilizing both the original and deformed depth images. Notably, our model achieves effective results with only a small dataset of approximately 300 samples. Experimental results on animal and human datasets demonstrate that our model outperforms other state-of-the-art methods.