Deng Jiacheng ; Wang Chuxin ; Lu Jiahao ; He Jianfeng ; Zhang Tianzhu ; Yu Jiyang ; Zhang Zhe

Abstract

Unsupervised point cloud shape correspondence aims to obtain densepoint-to-point correspondences between point clouds without manually annotatedpairs. However, humans and some animals have bilateral symmetry and variousorientations, which lead to severe mispredictions of symmetrical parts.Besides, point cloud noise disrupts consistent representations for point cloudand thus degrades the shape correspondence accuracy. To address the aboveissues, we propose a Self-Ensembling ORientation-aware Network termed SE-ORNet.The key of our approach is to exploit an orientation estimation module with adomain adaptive discriminator to align the orientations of point cloud pairs,which significantly alleviates the mispredictions of symmetrical parts.Additionally, we design a selfensembling framework for unsupervised point cloudshape correspondence. In this framework, the disturbances of point cloud noiseare overcome by perturbing the inputs of the student and teacher networks withdifferent data augmentations and constraining the consistency of predictions.Extensive experiments on both human and animal datasets show that our SE-ORNetcan surpass state-of-the-art unsupervised point cloud shape correspondencemethods.