3D Computer Vision

ASSANet: An Anisotropic Separable Set Abstraction for Efficient Point Cloud Representation Learning