CurveCloudNet: Processing Point Clouds with 1D Structure

TL;DR

CurveCloudNet is a novel point cloud processing backbone that leverages the inherent 1D curve-like structure of LiDAR data to improve segmentation accuracy and scalability across diverse 3D environments.

Contribution

It introduces CurveCloudNet, a new method that parameterizes point clouds as polylines to utilize local surface-aware ordering and curve reasoning, outperforming existing methods.

Findings

Outperforms point-based and sparse-voxel backbones in segmentation tasks.

Scales better to large scenes than point-based methods.

Shows improved single-object segmentation over sparse-voxel approaches.

Abstract

Modern depth sensors such as LiDAR operate by sweeping laser-beams across the scene, resulting in a point cloud with notable 1D curve-like structures. In this work, we introduce a new point cloud processing scheme and backbone, called CurveCloudNet, which takes advantage of the curve-like structure inherent to these sensors. While existing backbones discard the rich 1D traversal patterns and rely on generic 3D operations, CurveCloudNet parameterizes the point cloud as a collection of polylines (dubbed a "curve cloud"), establishing a local surface-aware ordering on the points. By reasoning along curves, CurveCloudNet captures lightweight curve-aware priors to efficiently and accurately reason in several diverse 3D environments. We evaluate CurveCloudNet on multiple synthetic and real datasets that exhibit distinct 3D size and structure. We demonstrate that CurveCloudNet outperforms both point-based and sparse-voxel backbones in various segmentation settings, notably scaling to large scenes better than point-based alternatives while exhibiting improved single-object performance over sparse-voxel alternatives. In all, CurveCloudNet is an efficient and accurate backbone that can handle a larger variety of 3D environments than past works.