Enhanced Prototypical Learning for Unsupervised Domain Adaptation in LiDAR Semantic Segmentation

TL;DR

This paper introduces an efficient range image-based method for unsupervised domain adaptation in LiDAR semantic segmentation, leveraging class prototypes and pre-training to improve performance without complex multi-stage inference.

Contribution

It proposes a novel prototypical learning approach with reconstruction pre-training and selective pseudo labeling tailored for LiDAR UDA, addressing domain shift challenges.

Findings

Achieves state-of-the-art performance on LiDAR UDA benchmarks.

Demonstrates effectiveness of prototype-based pseudo labeling in LiDAR segmentation.

Provides a real-time capable solution for LiDAR domain adaptation.

Abstract

Despite its importance, unsupervised domain adaptation (UDA) on LiDAR semantic segmentation is a task that has not received much attention from the research community. Only recently, a completion-based 3D method has been proposed to tackle the problem and formally set up the adaptive scenarios. However, the proposed pipeline is complex, voxel-based and requires multi-stage inference, which inhibits it for real-time inference. We propose a range image-based, effective and efficient method for solving UDA on LiDAR segmentation. The method exploits class prototypes from the source domain to pseudo label target domain pixels, which is a research direction showing good performance in UDA for natural image semantic segmentation. Applying such approaches to LiDAR scans has not been considered because of the severe domain shift and lack of pre-trained feature extractor that is unavailable in the LiDAR segmentation setup. However, we show that proper strategies, including reconstruction-based pre-training, enhanced prototypes, and selective pseudo labeling based on distance to prototypes, is sufficient enough to enable the use of prototypical approaches. We evaluate the performance of our method on the recently proposed LiDAR segmentation UDA scenarios. Our method achieves remarkable performance among contemporary methods.