PointMixer: MLP-Mixer for Point Cloud Understanding

TL;DR

PointMixer introduces a simple yet effective MLP-based operator for point cloud understanding, enabling competitive performance in various 3D recognition tasks by replacing token-mixing MLPs with a softmax function.

Contribution

It proposes PointMixer, a universal point set operator that facilitates feature mixing in unstructured 3D point clouds, broadening the application of MLP-Mixer concepts to 3D data.

Findings

Achieves competitive or superior results in semantic segmentation.

Outperforms transformer-based methods in classification tasks.

Effective in point reconstruction applications.

Abstract

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 visual recognition tasks. Unlike images, point clouds are inherently sparse, unordered and irregular, which limits the direct use of MLP-Mixer for point cloud understanding. In this paper, we propose PointMixer, a universal point set operator that facilitates information sharing among unstructured 3D points. By simply replacing token-mixing MLPs with a softmax function, PointMixer can "mix" features within/between point sets. By doing so, PointMixer can be broadly used in the network as inter-set mixing, intra-set mixing, and pyramid mixing. Extensive experiments show the competitive or superior performance of PointMixer in semantic segmentation, classification, and point reconstruction against transformer-based methods.