mlsanityData to 7 Oct 2026

Towards Efficient Point Cloud Graph Neural Networks Through Architectural Simplification

arXiv 2108.06317 ↗First posted 13 Aug 2021cs.CV cs.LGcited by 9

Not on the map. Why?

Abstract (arXiv)

In recent years graph neural network (GNN)-based approaches have become a popular strategy for processing point cloud data, regularly achieving state-of-the-art performance on a variety of tasks. To date, the research community has primarily focused on improving model expressiveness, with secondary thought given to how to design models that can run efficiently on resource constrained mobile devices including smartphones or mixed reality headsets. In this work we make a step towards improving the efficiency of these models by making the observation that these GNN models are heavily limited by the representational power of their first, feature extracting, layer. We find that it is possible to radically simplify these models so long as the feature extraction layer is retained with minimal degradation to model performance; further, we discover that it is possible to improve performance overall on ModelNet40 and S3DIS by improving the design of the feature extractor. Our approach reduces memory consumption by 20× and latency by up to 9.9× for graph layers in models such as DGCNN; overall, we achieve speed-ups of up to 4.5× and peak memory reductions of 72.5%.

Key facts. Cited by 9 papers in the map (data through 7 Oct 2026). Builds on 33 of its 40 references that are in the map. First posted on arXiv 13 Aug 2021.

Builds on

33 of 40 references.

References · 33

  1. 2017Attention Is All You Needcited by 59,932
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  3. 2016Semi-Supervised Classification with Graph Convolutional Networkscited by 11,282
  4. 2017Graph Attention Networkscited by 8,854
  5. 2016PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationcited by 5,223
  6. 2017PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Spacecited by 4,384
  7. 2018How Powerful are Graph Neural Networks?cited by 4,355
  8. 2017Neural Message Passing for Quantum Chemistrycited by 3,460
  9. 2019A Comprehensive Survey on Graph Neural Networkscited by 2,906
  10. 2018Dynamic Graph CNN for Learning on Point Cloudscited by 2,220
  11. 2020Conformer: Convolution-augmented Transformer for Speech Recognitioncited by 1,838
  12. 2020Rethinking Attention with Performerscited by 1,584
21 more
  1. 2020Point Transformercited by 1,462
  2. 2019KPConv: Flexible and Deformable Convolution for Point Cloudscited by 1,202
  3. 2018PointCNN: Convolution On X-Transformed Pointscited by 764
  4. 2018PointConv: Deep Convolutional Networks on 3D Point Cloudscited by 656
  5. 2019RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Cloudscited by 584
  6. 2015Multi-view Convolutional Neural Networks for 3D Shape Recognitioncited by 564
  7. 2020Principal Neighbourhood Aggregation for Graph Netscited by 477
  8. 2017Efficient Processing of Deep Neural Networks: A Tutorial and Surveycited by 427
  9. 2017Joint 2D-3D-Semantic Data for Indoor Scene Understandingcited by 404
  10. 2016OctNet: Learning Deep 3D Representations at High Resolutionscited by 340
  11. 2018SpiderCNN: Deep Learning on Point Sets with Parameterized Convolutional Filterscited by 261
  12. 2021PAConv: Position Adaptive Convolution with Dynamic Kernel Assembling on Point Cloudscited by 211
  13. 2021Revisiting Point Cloud Shape Classification with a Simple and Effective Baselinecited by 145
  14. 2020Kernel Operations on the GPU, with Autodiff, without Memory Overflowscited by 101
  15. 2020Degree-Quant: Quantization-Aware Training for Graph Neural Networkscited by 93
  16. 2020A Closer Look at Local Aggregation Operators in Point Cloud Analysiscited by 87
  17. 2020BiPointNet: Binary Neural Network for Point Cloudscited by 30
  18. 2018View-volume Network for Semantic Scene Completion from a Single Depth Imagecited by 26
  19. 2021Towards Efficient Graph Convolutional Networks for Point Cloud Handlingcited by 18
  20. 2020Probabilistic Dual Network Architecture Search on Graphscited by 15
  21. 2021Exploiting Local Geometry for Feature and Graph Construction for Better 3D Point Cloud Processing with Graph Neural Networkscited by 9

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2022: 1; 2023: 3; 2024: 3; 2025: 1; 2026: 1.

9 papers.

  1. 2023Mind the Label Shift of Augmentation-based Graph OOD Generalizationcited by 34
  2. 2022Parallel and Distributed Graph Neural Networks: An In-Depth Concurrency Analysiscited by 29
  3. 2023Hardware-Aware Graph Neural Network Automated Design for Edge Computing Platformscited by 12
  4. 2023Decoupled Local Aggregation for Point Cloud Learningcited by 12
  5. 2024LMSeg: An end-to-end geometric message-passing network on barycentric dual graphs for large-scale landscape mesh segmentationcited by 0
  6. 2024Learning Geometric Invariant Features for Classification of Vector Polygons with Graph Message-passing Neural Networkcited by 0
  7. 2024HGNAS: Hardware-Aware Graph Neural Architecture Search for Edge Devicescited by 0
  8. 2025GCoDE: Efficient Device-Edge Co-Inference for GNNs via Architecture-Mapping Co-Searchcited by 0
  9. 2026Rotation-Invariant Multi-IMU Activity Recognition under Independent Per-Location Orientation Shiftscited by 0