PointAtrousNet: Point Atrous Convolution for Point Cloud Analysis

Liang Pan, Pengfei Wang, Chee-Meng Chew

Abstract: In this letter, we propose a permutation-invariant architecture-PointAtrousNet (PAN), which focuses on exploiting multi-scale local geometric details for point cloud analysis. Inspired by atrous convolution in image domains, we propose the point atrous convolution (PAC) operation. Our PAC can effectively enlarge the receptive field of filters without introducing more parameters or increasing computation amount. In particular, we propose a novel point atrous spatial pyramid pooling module to explicitly exploit neighboring contextual information at multiple scales. Moreover, local geometric details are captured by constructing neighborhood graphs in metric and feature spaces. Experimental results show that our PAN achieves state-of-the-art performance on various point cloud inference applications.