LiDAR-UDA: Self-ensembling Through Time for Unsupervised LiDAR Domain Adaptation

TL;DR

LiDAR-UDA introduces a two-stage self-training approach for unsupervised LiDAR domain adaptation, utilizing beam subsampling and cross-frame ensembling to enhance pseudo label quality and reduce sensor discrepancy, leading to improved segmentation performance.

Contribution

The paper proposes novel techniques—LiDAR beam subsampling and cross-frame ensembling—to address sensor discrepancies in LiDAR UDA, improving pseudo label accuracy without extra inference costs.

Findings

Outperforms state-of-the-art by over 3.9% mIoU on multiple datasets

Effective in reducing domain shift caused by different LiDAR sensors

No additional inference cost introduced by the proposed methods

Abstract

We introduce LiDAR-UDA, a novel two-stage self-training-based Unsupervised Domain Adaptation (UDA) method for LiDAR segmentation. Existing self-training methods use a model trained on labeled source data to generate pseudo labels for target data and refine the predictions via fine-tuning the network on the pseudo labels. These methods suffer from domain shifts caused by different LiDAR sensor configurations in the source and target domains. We propose two techniques to reduce sensor discrepancy and improve pseudo label quality: 1) LiDAR beam subsampling, which simulates different LiDAR scanning patterns by randomly dropping beams; 2) cross-frame ensembling, which exploits temporal consistency of consecutive frames to generate more reliable pseudo labels. Our method is simple, generalizable, and does not incur any extra inference cost. We evaluate our method on several public LiDAR datasets and show that it outperforms the state-of-the-art methods by more than $3.9\%$ mIoU on average for all scenarios. Code will be available at https://github.com/JHLee0513/LiDARUDA.