SPNet: Multi-Shell Kernel Convolution for Point Cloud Semantic Segmentation

TL;DR

This paper introduces SPNet, a novel neural network utilizing Shell Point Convolution (SPConv) with shell-based feature encoding and attention mechanisms, achieving top performance in large-scale point cloud semantic segmentation.

Contribution

The paper proposes SPConv, a new point convolution operator that splits neighborhoods into shells and uses attention, and integrates it into SPNet for improved point cloud segmentation.

Findings

SPConv effectively encodes local shape features.

SPNet achieves top-ranking results on large-scale datasets.

Shell design and attention improve segmentation accuracy.

Abstract

Feature encoding is essential for point cloud analysis. In this paper, we propose a novel point convolution operator named Shell Point Convolution (SPConv) for shape encoding and local context learning. Specifically, SPConv splits 3D neighborhood space into shells, aggregates local features on manually designed kernel points, and performs convolution on the shells. Moreover, SPConv incorporates a simple yet effective attention module that enhances local feature aggregation. Based upon SPConv, a deep neural network named SPNet is constructed to process large-scale point clouds. Poisson disk sampling and feature propagation are incorporated in SPNet for better efficiency and accuracy. We provided details of the shell design and conducted extensive experiments on challenging large-scale point cloud datasets. Experimental results show that SPConv is effective in local shape encoding, and our SPNet is able to achieve top-ranking performances in semantic segmentation tasks.