Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Correspondence Learning via Linearly-invariant Embedding

About

In this paper, we propose a fully differentiable pipeline for estimating accurate dense correspondences between 3D point clouds. The proposed pipeline is an extension and a generalization of the functional maps framework. However, instead of using the Laplace-Beltrami eigenfunctions as done in virtually all previous works in this domain, we demonstrate that learning the basis from data can both improve robustness and lead to better accuracy in challenging settings. We interpret the basis as a learned embedding into a higher dimensional space. Following the functional map paradigm the optimal transformation in this embedding space must be linear and we propose a separate architecture aimed at estimating the transformation by learning optimal descriptor functions. This leads to the first end-to-end trainable functional map-based correspondence approach in which both the basis and the descriptors are learned from data. Interestingly, we also observe that learning a \emph{canonical} embedding leads to worse results, suggesting that leaving an extra linear degree of freedom to the embedding network gives it more robustness, thereby also shedding light onto the success of previous methods. Finally, we demonstrate that our approach achieves state-of-the-art results in challenging non-rigid 3D point cloud correspondence applications.

Riccardo Marin, Marie-Julie Rakotosaona, Simone Melzi, Maks Ovsjanikov• 2020

Related benchmarks

TaskDatasetResultRank
3D Shape CorrespondenceFAUST remeshed (test)
Mean Geodesic Error (x100)3.6
65
Shape MatchingSCAPE remeshed (test)
Mean Geodesic Error (x100)12
46
Non-rigid shape matchingDT4D-H
Mean Geodesic Error (x100)15.9
39
Shape MatchingSHREC19 remeshed (test)
Mean Geodesic Error0.164
37
Near-isometric point cloud matchingSCAPE_r remeshed (test)
Mean Geodesic Error0.12
25
Point cloud matchingFAUST_r
Mean Geodesic Error0.036
23
Point cloud matchingSCAPE_r
Mean Geodesic Error12
23
Non-isometric 3D shape matchingSMAL
Mean Geodesic Error20
22
Non-rigid shape matchingSHREC H '07
Mean Geodesic Error0.154
20
Shape CorrespondenceSurreal (test)
Accuracy4
16
Showing 10 of 24 rows

Other info

Follow for update