PointVector: A Vector Representation In Point Cloud Analysis

TL;DR

PointVector introduces a vector-based feature aggregation method for point cloud analysis, enhancing local feature extraction and achieving state-of-the-art results with fewer parameters.

Contribution

The paper proposes a novel vector-oriented point set abstraction and a transformation from scalar to vector features, improving local feature extraction in point cloud analysis.

Findings

Achieves 72.3% mIOU on S3DIS Area 5

Achieves 78.4% mIOU on S3DIS cross-validation

Uses only 58% of PointNeXt's parameters

Abstract

In point cloud analysis, point-based methods have rapidly developed in recent years. These methods have recently focused on concise MLP structures, such as PointNeXt, which have demonstrated competitiveness with Convolutional and Transformer structures. However, standard MLPs are limited in their ability to extract local features effectively. To address this limitation, we propose a Vector-oriented Point Set Abstraction that can aggregate neighboring features through higher-dimensional vectors. To facilitate network optimization, we construct a transformation from scalar to vector using independent angles based on 3D vector rotations. Finally, we develop a PointVector model that follows the structure of PointNeXt. Our experimental results demonstrate that PointVector achieves state-of-the-art performance $\textbf{72.3\% mIOU}$ on the S3DIS Area 5 and $\textbf{78.4\% mIOU}$ on the S3DIS (6-fold cross-validation) with only $\textbf{58\%}$ model parameters of PointNeXt. We hope our work will help the exploration of concise and effective feature representations. The code will be released soon.