LiDAR-UDA: Self-ensembling Through Time for Unsupervised LiDAR Domain Adaptation
This work addresses domain adaptation for LiDAR segmentation, which is crucial for autonomous driving systems, but it is incremental as it builds on existing self-training methods with specific enhancements.
The paper tackles the problem of domain shift in LiDAR segmentation by introducing LiDAR-UDA, a self-training-based unsupervised domain adaptation method that uses LiDAR beam subsampling and cross-frame ensembling to improve pseudo label quality, achieving an average improvement of over 3.9% mIoU compared to state-of-the-art methods.
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.