Instance Segmentation for Point Sets

TL;DR

This paper introduces sampling-based methods for 3D point set instance segmentation that reduce memory usage and improve speed by extrapolating labels from sub-sampled data, addressing limitations of existing neural network approaches.

Contribution

It proposes two novel sampling strategies to perform instance segmentation on point clouds, significantly reducing memory and computational requirements compared to previous methods.

Findings

Random sampling approach improves speed and memory efficiency.

Both sampling methods perform equally well on large sub-samples.

The approach enables scalable instance segmentation for large point sets.

Abstract

Recently proposed neural network architectures like PointNet [QSMG16] and PointNet++ [QYSG17] have made it possible to apply Deep Learning to 3D point sets. The feature representations of shapes learned by these two networks enabled training classifiers for Semantic Segmentation, and more recently for Instance Segmentation via the Similarity Group Proposal Network (SGPN) [WYHN17]. One area of improvement which has been highlighted by SGPN's authors, pertains to use of memory intensive similarity matrices which occupy memory quadratic in the number of points. In this report, we attempt to tackle this issue through use of two sampling based methods, which compute Instance Segmentation on a sub-sampled Point Set, and then extrapolate labels to the complete set using the nearest neigbhour approach. While both approaches perform equally well on large sub-samples, the random-based strategy gives the most improvements in terms of speed and memory usage.