PointMatch: A consistency training framework for weakly supervised semantic segmentation of 3D point clouds

Yushuang Wu, Zizheng Yan, Shengcai Cai, Guanbin Li, Xiaoguang Han, Shuguang Cui

Abstract: Highlights•A consistency training framework for efficient learning from sparse semantic labels.•Propose to probe and well exploit super-points to promote the pseudo-label quality.•Extensive experiments validate the effectiveness and superiority of PointMatch.