GeoAuxNet: Towards Universal 3D Representation Learning for Multi-sensor Point Clouds

arXiv:2403.19220v1 Announce Type: new
Abstract: Point clouds captured by different sensors such as RGB-D cameras and LiDAR possess non-negligible domain gaps. Most existing methods design different network architectures and train separately on point clouds from various sensors. Typically, point-based methods achieve outstanding performances on even-distributed dense point clouds from RGB-D cameras, while voxel-based methods are more efficient for large-range sparse LiDAR point clouds. In this paper, we propose geometry-to-voxel auxiliary learning to enable voxel representations to access point-level geometric information, …

abstract architectures arxiv cameras cs.cv design distributed domain lidar network performances representation representation learning rgb-d sensor sensors train type universal