LMSCNet: Lightweight Multiscale 3D Semantic Completion

TL;DR

LMSCNet is a lightweight, multiscale 3D scene completion method using a 2D UNet backbone that achieves competitive accuracy with higher speed and efficiency, ideal for mobile robotics.

Contribution

It introduces a novel multiscale 3D semantic completion approach with a 2D UNet backbone and skip connections, outperforming existing methods in speed and efficiency.

Findings

Performs on par with state-of-the-art in semantic completion

Outperforms others in occupancy completion

Offers high speed at coarse levels

Abstract

We introduce a new approach for multiscale 3Dsemantic scene completion from voxelized sparse 3D LiDAR scans. As opposed to the literature, we use a 2D UNet backbone with comprehensive multiscale skip connections to enhance feature flow, along with 3D segmentation heads. On the SemanticKITTI benchmark, our method performs on par on semantic completion and better on occupancy completion than all other published methods -- while being significantly lighter and faster. As such it provides a great performance/speed trade-off for mobile-robotics applications. The ablation studies demonstrate our method is robust to lower density inputs, and that it enables very high speed semantic completion at the coarsest level. Our code is available at https://github.com/cv-rits/LMSCNet.