资讯

Abstract: Masked Autoencoders (MAE) demonstrated exceptional performance in natural language processing and 2D vision tasks and have now been introduced into point cloud representation learning. We ...