MA-ST3D: Motion Associated Self-Training for Unsupervised Domain Adaptation on 3D Object Detection

Abstract

This paper addresses the problem of unsupervised domain adaptation for 3D object detection, where we aim to adapt a detector trained on a source domain to perform well on a target domain without any labeled data. We propose MA-ST3D, a novel method that leverages motion association to improve the self-training process.

BibTeX

@article{zhang2024mast3d,
  title={MA-ST3D: Motion Associated Self-Training for Unsupervised Domain Adaptation on 3D Object Detection},
  author={Zhang, Chi and Chen, Wenbo and Wang, Wei and Zhang, Zhaoxiang},
  journal={IEEE Transactions on Image Processing},
  volume={33},
  pages={3980--3993},
  year={2024},
  publisher={IEEE}
}