DPFM: Deep Partial Functional Maps

TL;DR

This paper introduces DPFM, a deep learning-based method for dense partial non-rigid shape correspondence that learns descriptors directly from data, improving robustness and accuracy over traditional spectral methods.

Contribution

It presents the first learning-based approach for partial non-rigid shape correspondence using the functional map framework, capable of partial-to-partial matching without prior common region knowledge.

Findings

Achieves state-of-the-art results on benchmark datasets.

Data-efficient and applicable to partial-to-partial matching.

Improves robustness and accuracy over existing spectral methods.

Abstract

We consider the problem of computing dense correspondences between non-rigid shapes with potentially significant partiality. Existing formulations tackle this problem through heavy manifold optimization in the spectral domain, given hand-crafted shape descriptors. In this paper, we propose the first learning method aimed directly at partial non-rigid shape correspondence. Our approach uses the functional map framework, can be trained in a supervised or unsupervised manner, and learns descriptors directly from the data, thus both improving robustness and accuracy in challenging cases. Furthermore, unlike existing techniques, our method is also applicable to partial-to-partial non-rigid matching, in which the common regions on both shapes are unknown a priori. We demonstrate that the resulting method is data-efficient, and achieves state-of-the-art results on several benchmark datasets. Our code and data can be found online: https://github.com/pvnieo/DPFM