FastPoint: Accelerating 3D Point Cloud Model Inference via Sample Point Distance Prediction
Not on the map. Why?
Abstract (arXiv)
Deep neural networks have revolutionized 3D point cloud processing, yet efficiently handling large and irregular point clouds remains challenging. To tackle this problem, we introduce FastPoint, a novel software-based acceleration technique that leverages the predictable distance trend between sampled points during farthest point sampling. By predicting the distance curve, we can efficiently identify subsequent sample points without exhaustively computing all pairwise distances. Our proposal substantially accelerates farthest point sampling and neighbor search operations while preserving sampling quality and model performance. By integrating FastPoint into state-of-the-art 3D point cloud models, we achieve 2.55x end-to-end speedup on NVIDIA RTX 3090 GPU without sacrificing accuracy.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 33 of its 45 references that are in the map. First posted on arXiv 31 Jul 2025.
Builds on
33 of 45 references.
References · 33
- 2016PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationcited by 5,248
- 2017PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Spacecited by 4,401
- 2017ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenescited by 3,117
- 2018Dynamic Graph CNN for Learning on Point Cloudscited by 2,223
- 2020Point Transformercited by 1,470
- 20194D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networkscited by 1,208
- 2019KPConv: Flexible and Deformable Convolution for Point Cloudscited by 1,205
- 2018PointConv: Deep Convolutional Networks on 3D Point Cloudscited by 657
- 2017Frustum PointNets for 3D Object Detection from RGB-D Datacited by 640
- 2023Point Transformer V3: Simpler, Faster, Strongercited by 589
- 2019RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Cloudscited by 585
- 2022PointNeXt: Revisiting PointNet++ with Improved Training and Scaling Strategiescited by 436
21 more
- 2020Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR Segmentationcited by 389
- 2022Point Transformer V2: Grouped Vector Attention and Partition-based Poolingcited by 388
- 2021Voxel Transformer for 3D Object Detectioncited by 247
- 2021PAConv: Position Adaptive Convolution with Dynamic Kernel Assembling on Point Cloudscited by 211
- 2021Embracing Single Stride 3D Object Detector with Sparse Transformercited by 193
- 2020OccuSeg: Occupancy-aware 3D Instance Segmentationcited by 177
- 2024PointMamba: A Simple State Space Model for Point Cloud Analysiscited by 167
- 2021ImVoxelNet: Image to Voxels Projection for Monocular and Multi-View General-Purpose 3D Object Detectioncited by 157
- 2021Mix3D: Out-of-Context Data Augmentation for 3D Scenescited by 126
- 2023Swin3D: A Pretrained Transformer Backbone for 3D Indoor Scene Understandingcited by 113
- 2019DensePoint: Learning Densely Contextual Representation for Efficient Point Cloud Processingcited by 104
- 2019Grid-GCN for Fast and Scalable Point Cloud Learningcited by 103
- 2023FlatFormer: Flattened Window Attention for Efficient Point Cloud Transformercited by 73
- 2022LargeKernel3D: Scaling up Kernels in 3D Sparse CNNscited by 72
- 2024Point Cloud Mamba: Point Cloud Learning via State Space Modelcited by 66
- 2021ASSANet: An Anisotropic Separable Set Abstraction for Efficient Point Cloud Representation Learningcited by 29
- 2022PointVector: A Vector Representation In Point Cloud Analysiscited by 21
- 2022An Adjustable Farthest Point Sampling Method for Approximately-sorted Point Cloud Datacited by 14
- 2020Mesorasi: Architecture Support for Point Cloud Analytics via Delayed-Aggregationcited by 11
- 2022Crescent: Taming Memory Irregularities for Accelerating Deep Point Cloud Analyticscited by 8
- 2024Swin3D++: Effective Multi-Source Pretraining for 3D Indoor Scene Understandingcited by 6
Built on by
Citing papers per year: 2026: 1.
1 paper.