WSDesc: Weakly Supervised 3D Local Descriptor Learning for Point Cloud Registration

TL;DR

This paper introduces WSDesc, a weakly supervised method for learning 3D local descriptors that adaptively optimize local support size, improving point cloud registration without requiring ground-truth alignments.

Contribution

The work proposes a differentiable voxelization layer and a novel registration loss to learn 3D descriptors with weak supervision, enhancing registration accuracy.

Findings

Outperforms existing descriptors on geometric registration benchmarks.

Learns optimal local support size in a data-driven manner.

Does not require ground-truth alignment for training.

Abstract

In this work, we present a novel method called WSDesc to learn 3D local descriptors in a weakly supervised manner for robust point cloud registration. Our work builds upon recent 3D CNN-based descriptor extractors, which leverage a voxel-based representation to parameterize local geometry of 3D points. Instead of using a predefined fixed-size local support in voxelization, we propose to learn the optimal support in a data-driven manner. To this end, we design a novel differentiable voxelization layer that can back-propagate the gradient to the support size optimization. To train the extracted descriptors, we propose a novel registration loss based on the deviation from rigidity of 3D transformations, and the loss is weakly supervised by the prior knowledge that the input point clouds have partial overlap, without requiring ground-truth alignment information. Through extensive experiments, we show that our learned descriptors yield superior performance on existing geometric registration benchmarks.