Semantic Segmentation on 3D Point Clouds with High Density Variations

TL;DR

This paper introduces HDVNet, a novel neural network architecture designed to improve semantic segmentation of large-scale 3D point clouds with significant local density variations, common in LiDAR surveying data.

Contribution

HDVNet employs nested encoder-decoder pathways for different density ranges, effectively handling input density variations and outperforming existing models in accuracy with fewer parameters.

Findings

HDVNet achieves higher segmentation accuracy on real LiDAR point clouds.

HDVNet uses fewer weights than state-of-the-art models.

The architecture effectively manages large local density variations.

Abstract

LiDAR scanning for surveying applications acquire measurements over wide areas and long distances, which produces large-scale 3D point clouds with significant local density variations. While existing 3D semantic segmentation models conduct downsampling and upsampling to build robustness against varying point densities, they are less effective under the large local density variations characteristic of point clouds from surveying applications. To alleviate this weakness, we propose a novel architecture called HDVNet that contains a nested set of encoder-decoder pathways, each handling a specific point density range. Limiting the interconnections between the feature maps enables HDVNet to gauge the reliability of each feature based on the density of a point, e.g., downweighting high density features not existing in low density objects. By effectively handling input density variations, HDVNet outperforms state-of-the-art models in segmentation accuracy on real point clouds with inconsistent density, using just over half the weights.