3DSFLabelling: Boosting 3D Scene Flow Estimation by Pseudo Auto-labelling

arXiv:2402.18146v1 Announce Type: new
Abstract: Learning 3D scene flow from LiDAR point clouds presents significant difficulties, including poor generalization from synthetic datasets to real scenes, scarcity of real-world 3D labels, and poor performance on real sparse LiDAR point clouds. We present a novel approach from the perspective of auto-labelling, aiming to generate a large number of 3D scene flow pseudo labels for real-world LiDAR point clouds. Specifically, we employ the assumption of rigid body motion to simulate potential object-level rigid …

abstract arxiv auto boosting cs.cv datasets flow labelling labels lidar novel performance perspective synthetic type world