mlsanityData to 7 Oct 2026

PAConv: Position Adaptive Convolution with Dynamic Kernel Assembling on Point Clouds

arXiv 2103.14635 ↗First posted 26 Mar 2021cs.CVcited by 211

Not on the map. Why?

Abstract (arXiv)

We introduce Position Adaptive Convolution (PAConv), a generic convolution operation for 3D point cloud processing. The key of PAConv is to construct the convolution kernel by dynamically assembling basic weight matrices stored in Weight Bank, where the coefficients of these weight matrices are self-adaptively learned from point positions through ScoreNet. In this way, the kernel is built in a data-driven manner, endowing PAConv with more flexibility than 2D convolutions to better handle the irregular and unordered point cloud data. Besides, the complexity of the learning process is reduced by combining weight matrices instead of brutally predicting kernels from point positions. Furthermore, different from the existing point convolution operators whose network architectures are often heavily engineered, we integrate our PAConv into classical MLP-based point cloud pipelines without changing network configurations. Even built on simple networks, our method still approaches or even surpasses the state-of-the-art models, and significantly improves baseline performance on both classification and segmentation tasks, yet with decent efficiency. Thorough ablation studies and visualizations are provided to understand PAConv. Code is released on https://github.com/CVMI-Lab/PAConv.

Key facts. Cited by 211 papers in the map (data through 7 Oct 2026). Builds on 47 of its 65 references that are in the map. First posted on arXiv 26 Mar 2021.

Builds on

47 of 65 references.

References · 47

  1. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  2. 2016PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationcited by 5,223
  3. 2017PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Spacecited by 4,384
  4. 2018Dynamic Graph CNN for Learning on Point Cloudscited by 2,220
  5. 2019KPConv: Flexible and Deformable Convolution for Point Cloudscited by 1,202
  6. 20194D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networkscited by 1,201
  7. 2018PointCNN: Convolution On X-Transformed Pointscited by 764
  8. 20173D Semantic Segmentation with Submanifold Sparse Convolutional Networkscited by 746
  9. 2018PointConv: Deep Convolutional Networks on 3D Point Cloudscited by 656
  10. 2017Frustum PointNets for 3D Object Detection from RGB-D Datacited by 640
  11. 2019RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Cloudscited by 584
  12. 2015Multi-view Convolutional Neural Networks for 3D Shape Recognitioncited by 564
35 more
  1. 2019Relation-Shape Convolutional Neural Network for Point Cloud Analysiscited by 346
  2. 2019Point-Voxel CNN for Efficient 3D Deep Learningcited by 343
  3. 2016OctNet: Learning Deep 3D Representations at High Resolutionscited by 340
  4. 2016Dynamic Filter Networkscited by 318
  5. 2017Dynamic Edge-Conditioned Filters in Convolutional Neural Networks on Graphscited by 307
  6. 2019CondConv: Conditionally Parameterized Convolutions for Efficient Inferencecited by 273
  7. 2018SpiderCNN: Deep Learning on Point Sets with Parameterized Convolutional Filterscited by 261
  8. 2018SO-Net: Self-Organizing Network for Point Cloud Analysiscited by 210
  9. 2017SEGCloud: Semantic Segmentation of 3D Point Cloudscited by 178
  10. 2018Tangent Convolutions for Dense Prediction in 3Dcited by 173
  11. 2021Deep Parametric Continuous Convolutional Neural Networkscited by 161
  12. 2018SPLATNet: Sparse Lattice Networks for Point Cloud Processingcited by 158
  13. 2018Point Convolutional Neural Networks by Extension Operatorscited by 154
  14. 2017Escape from Cells: Deep Kd-Networks for the Recognition of 3D Point Cloud Modelscited by 142
  15. 2017Pointwise Convolutional Neural Networkscited by 126
  16. 2019DensePoint: Learning Densely Contextual Representation for Efficient Point Cloud Processingcited by 104
  17. 2019Grid-GCN for Fast and Scalable Point Cloud Learningcited by 103
  18. 2018Point2Sequence: Learning the Shape Representation of 3D Point Clouds with an Attention-based Sequence to Sequence Networkcited by 100
  19. 2020A Closer Look at Local Aggregation Operators in Point Cloud Analysiscited by 87
  20. 2019Interpolated Convolutional Networks for 3D Point Cloud Understandingcited by 86
  21. 2018Local Spectral Graph Convolution for Point Set Feature Learningcited by 84
  22. 2018Recurrent Slice Networks for 3D Segmentation of Point Cloudscited by 82
  23. 2017Deep Projective 3D Semantic Segmentationcited by 81
  24. 2020Learning Geometry-Disentangled Representation for Complementary Understanding of 3D Object Point Cloudcited by 71
  25. 2016SyncSpecCNN: Synchronized Spectral CNN for 3D Shape Segmentationcited by 70
  26. 2019Hierarchical Point-Edge Interaction Network for Point Cloud Semantic Segmentationcited by 68
  27. 2019A-CNN: Annularly Convolutional Neural Networks on Point Cloudscited by 66
  28. 2018VV-Net: Voxel VAE Net with Group Convolutions for Point Cloud Segmentationcited by 64
  29. 2019ConvPoint: Continuous Convolutions for Point Cloud Processingcited by 62
  30. 2020FPConv: Learning Local Flattening for Point Convolutioncited by 55
  31. 2019Geometry Sharing Network for 3D Point Cloud Classification and Segmentationcited by 33
  32. 2019Deformable Kernels: Adapting Effective Receptive Fields for Object Deformationcited by 29
  33. 2019Point2Node: Correlation Learning of Dynamic-Node for Point Cloud Feature Modelingcited by 25
  34. 2018Flex-Convolution (Million-Scale Point-Cloud Learning Beyond Grid-Worlds)cited by 23
  35. 2021Investigate Indistinguishable Points in Semantic Segmentation of 3D Point Cloudcited by 6

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 27; 2022: 53; 2023: 55; 2024: 32; 2025: 31; 2026: 13.

211 papers.

Sort:

By area:

  1. 2023Point Transformer V3: Simpler, Faster, Strongercited by 579
  2. 2022Point Transformer V2: Grouped Vector Attention and Partition-based Poolingcited by 384
  3. 2022Rethinking Network Design and Local Geometry in Point Cloud: A Simple Residual MLP Frameworkcited by 376
  4. 2023OmniObject3D: Large-Vocabulary 3D Object Dataset for Realistic Perception, Reconstruction and Generationcited by 324
  5. 2023MVImgNet: A Large-scale Dataset of Multi-view Imagescited by 219
  6. 2022CrossPoint: Self-Supervised Cross-Modal Contrastive Learning for 3D Point Cloud Understandingcited by 198
  7. 2022PLA: Language-Driven Open-Vocabulary 3D Scene Understandingcited by 176
  8. 2021Walk in the Cloud: Learning Curves for Point Clouds Shape Analysiscited by 172
  9. 2022Stratified Transformer for 3D Point Cloud Segmentationcited by 166
  10. 2022Learning 3D Representations from 2D Pre-trained Models via Image-to-Point Masked Autoencoderscited by 143
  11. 2025Sonata: Self-Supervised Learning of Reliable Point Representationscited by 108
  12. 2021Fast Point Transformercited by 103
38 more
  1. 2023OneFormer3D: One Transformer for Unified Point Cloud Segmentationcited by 95
  2. 2023RegionPLC: Regional Point-Language Contrastive Learning for Open-World 3D Scene Understandingcited by 93
  3. 2023Towards Large-scale 3D Representation Learning with Multi-dataset Point Prompt Trainingcited by 85
  4. 2022Surface Representation for Point Cloudscited by 84
  5. 2023CLIP goes 3D: Leveraging Prompt Tuning for Language Grounded 3D Recognitioncited by 77
  6. 2021PointMixer: MLP-Mixer for Point Cloud Understandingcited by 71
  7. 2022Dual Octree Graph Networks for Learning Adaptive Volumetric Shape Representationscited by 71
  8. 2022Benchmarking and Analyzing Point Cloud Classification under Corruptionscited by 70
  9. 2024Point Cloud Mamba: Point Cloud Learning via State Space Modelcited by 66
  10. 2023Deep Learning on Implicit Neural Representations of Shapescited by 62
  11. 2025DeepMesh: Auto-Regressive Artist-mesh Creation with Reinforcement Learningcited by 62
  12. 2022A Survey on Graph Neural Networks and Graph Transformers in Computer Vision: A Task-Oriented Perspectivecited by 56
  13. 2022Shape-invariant 3D Adversarial Point Cloudscited by 49
  14. 2023SAMPro3D: Locating SAM Prompts in 3D for Zero-Shot Instance Segmentationcited by 44
  15. 2024OA-CNNs: Omni-Adaptive Sparse CNNs for 3D Semantic Segmentationcited by 43
  16. 2023Self-positioning Point-based Transformer for Point Cloud Understandingcited by 40
  17. 2023Smooth, exact rotational symmetrization for deep learning on point cloudscited by 39
  18. 2023Lowis3D: Language-Driven Open-World Instance-Level 3D Scene Understandingcited by 38
  19. 2023Attention-based Point Cloud Edge Samplingcited by 37
  20. 2023Parameter is Not All You Need: Starting from Non-Parametric Networks for 3D Point Cloud Analysiscited by 34
  21. 2022A Unified Query-based Paradigm for Point Cloud Understandingcited by 32
  22. 2022Transformers in 3D Point Clouds: A Surveycited by 31
  23. 2021Dual Transformer for Point Cloud Analysiscited by 26
  24. 2022Isometric 3D Adversarial Examples in the Physical Worldcited by 25
  25. 2023Building3D: An Urban-Scale Dataset and Benchmarks for Learning Roof Structures from Point Cloudscited by 24
  26. 20213D Medical Point Transformer: Introducing Convolution to Attention Networks for Medical Point Cloud Analysiscited by 23
  27. 2022Bidirectional Feature Globalization for Few-shot Semantic Segmentation of 3D Point Cloud Scenescited by 23
  28. 2021Not All Voxels Are Equal: Semantic Scene Completion from the Point-Voxel Perspectivecited by 22
  29. 2022PointVector: A Vector Representation In Point Cloud Analysiscited by 21
  30. 2022FBNet: Feedback Network for Point Cloud Completioncited by 20
  31. 2022PeRFception: Perception using Radiance Fieldscited by 19
  32. 2023Point Cloud Classification Using Content-based Transformer via Clustering in Feature Spacecited by 19
  33. 2023MarS3D: A Plug-and-Play Motion-Aware Model for Semantic Segmentation on Multi-Scan 3D Point Cloudscited by 19
  34. 2021Do We Need Anisotropic Graph Neural Networks?cited by 18
  35. 2022DArch: Dental Arch Prior-assisted 3D Tooth Instance Segmentationcited by 18
  36. 2022The Devil is in the Pose: Ambiguity-free 3D Rotation-invariant Learning via Pose-aware Convolutioncited by 18
  37. 2022Rethinking the compositionality of point clouds through regularization in the hyperbolic spacecited by 18
  38. 2024PartSTAD: 2D-to-3D Part Segmentation Task Adaptationcited by 18

50 most cited of 211 All 211 (JSON)