Rethinking Network Design and Local Geometry in Point Cloud: A Simple Residual MLP Framework

By Xu Ma, Can Qin, Haoxuan You, Haoxi Ran, Yun Raymond Fu. International Conference on Learning Representations, 2022

913 citations recorded by the source corpus.

Point cloud analysis is challenging due to irregularity and unordered data structure. To capture the 3D geometries, prior works mainly rely on exploring sophisticated local geometric extractors using convolution, graph, or attention mechanisms. These methods, however, incur unfavorable latency during inference, and the performance saturates over the past few years. In this paper, we present a novel perspective on this task. We notice that detailed local geometrical information probably is not the key to point cloud analysis -- we introduce a pure residual MLP network, called PointMLP, which integrates no sophisticated local geometrical extractors but still performs very competitively. Equipped with a proposed lightweight geometric affine module, PointMLP delivers the new state-of-the-art on multiple datasets. On the real-world ScanObjectNN dataset, our method even surpasses the prior best method by 3.3% accuracy. We emphasize that PointMLP achieves this strong performance without any sophisticated operations, hence leading to a superior inference speed. Compared to most recent CurveNet, PointMLP trains 2x faster, tests 7x faster, and is more accurate on ModelNet40 benchmark. We hope our PointMLP may help the community towards a better understanding of point cloud analysis. The code is available at https://github.com/ma-xu/pointMLP-pytorch.

Subfields: Computational Mechanics, Environmental Engineering, Geology

View Paper: https://www.semanticscholar.org/paper/a7cc9851d78bd718e17f6fca05efa16710344952

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Mechanics

1. Xiaoyang Wu
2. Xihui Liu
3. Guocheng Qian
4. Zhikang Zou
5. Zhe Gan
6. Hasan Hammoud
7. Jinjie Mai
8. Mohamed Elhoseiny

In Environmental Engineering

1. Xiaoyang Wu
2. Xihui Liu
3. Hengshuang Zhao
4. Zhijian Liu

In Geology

1. Xiaoyang Wu
2. Xihui Liu
3. Guocheng Qian
4. Hasan Hammoud
5. Jinjie Mai
6. Mohamed Elhoseiny
7. Houwen Peng
8. Bernard Ghanem

Citations in the corpus

• Aligning Cyber Space With Physical World: A Comprehensive Survey on Embodied AI (2024, neutral): “Incontrast, point-basedmethods process point clouds directly (e.g., PointNet [75], PointNet++[76], PointMLP [77]).”
• Ferret: Refer and Ground Anything Anywhere at Any Granularity (2023, methodology): “Inspired by existing works in 3D point cloud learning (Qi et al., 2017a; Ma et al., 2022; Wang et al., 2019), we propose a spatial-aware visual sampler.”
• PointNeXt: Revisiting PointNet++ with Improved Training and Scaling Strategies (2022, methodology): “The improved training strategies are general and can be easily applied to improve other methods [29, 46, 28].”
• PointMamba: A Simple State Space Model for Point Cloud Analysis (2024, methodology): “Point Cloud Mamba (PCM) [63] combines an improved Mamba module, i.e., Vision Mamba [68], with PointMLP [37] (a strong point cloud analysis method), and incorporates consistent traverse serialization at each stage.”
• Point Transformer V3: Simpler, Faster, Stronger (2023, background): “Point-based methods, by contrast, process point clouds directly [52, 62, 63, 77, 105] and have recently seen a shift towards transformer-based architectures [27, 65, 90, 101, 106].”