Skip to Main content Skip to Navigation
Conference papers

SF-UDA$^{3D}$: Source-Free Unsupervised Domain Adaptation for LiDAR-Based 3D Object Detection

Abstract : 3D object detectors based only on LiDAR point clouds hold the state-of-the-art on modern street-view benchmarks. However, LiDAR-based detectors poorly generalize across domains due to domain shift. In the case of LiDAR, in fact, domain shift is not only due to changes in the environment and in the object appearances, as for visual data from RGB cameras, but is also related to the geometry of the point clouds (e.g., point density variations). This paper proposes SF-UDA$^{3D}$, the first Source-Free Unsupervised Domain Adaptation (SF-UDA) framework to domain-adapt the state-of-the-art PointRCNN 3D detector to target domains for which we have no annotations (unsupervised), neither we hold images nor annotations of the source domain (source-free). SF-UDA$^{3D}$ is novel on both aspects. Our approach is based on pseudo-annotations, reversible scale-transformations and motion coherency. SF-UDA$^{3D}$ outperforms both previous domain adaptation techniques based on features alignment and state-of-the-art 3D object detection methods which additionally use few-shot target annotations or target annotation statistics. This is demonstrated by extensive experiments on two large-scale datasets, i.e., KITTI and nuScenes.
Document type :
Conference papers
Complete list of metadata
Contributor : Stéphane Lathuilière Connect in order to contact the contributor
Submitted on : Wednesday, January 13, 2021 - 10:39:39 AM
Last modification on : Wednesday, November 3, 2021 - 2:24:42 PM

Links full text


  • HAL Id : hal-03108401, version 1
  • ARXIV : 2010.08243



Cristiano Saltori, Stéphane Lathuilière, Nicu Sebe, Elisa Ricci, Fabio Galasso. SF-UDA$^{3D}$: Source-Free Unsupervised Domain Adaptation for LiDAR-Based 3D Object Detection. International Conference on 3D Vision, Nov 2020, Fukuoka, Japan. ⟨hal-03108401⟩



Record views