PointMixer: MLP-Mixer for Point Cloud Understanding

• Published in: Computer Vision - ECCV 2022 Vol. 13687; pp. 620 - 640
• Main Authors: Choe, Jaesung, Park, Chunghyun, Rameau, Francois, Park, Jaesik, Kweon, In So
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer 01.01.2022; Springer Nature Switzerland
• Series: Lecture Notes in Computer Science
• ISBN: 9783031198113; 3031198115
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-031-19812-0_36

MLP-Mixer has newly appeared as a new challenger against the realm of CNNs and Transformer. Despite its simplicity compared to Transformer, the concept of channel-mixing MLPs and token-mixing MLPs achieves noticeable performance in image recognition tasks. Unlike images, point clouds are inherently sparse, unordered and irregular, which limits the direct use of MLP-Mixer for point cloud understanding. To overcome these limitations, we propose PointMixer, a universal point set operator that facilitates information sharing among unstructured 3D point cloud. By simply replacing token-mixing MLPs with Softmax function, PointMixer can “mix” features within/between point sets. By doing so, PointMixer can be broadly used for intra-set, inter-set, and hierarchical-set mixing. We demonstrate that various channel-wise feature aggregation in numerous point sets is better than self-attention layers or dense token-wise interaction in a view of parameter efficiency and accuracy. Extensive experiments show the competitive or superior performance of PointMixer in semantic segmentation, classification, and reconstruction against Transformer-based methods.