Research map / Point clouds / Point cloud classification
Mining Point Cloud Local Structures by Kernel Correlation and Graph Pooling
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add kernel correlation in place of handcrafted point normals (ablation) | KCNet · ModelNet40 | accuracy (%): 88.4 -> 90.5 | Helped |
| use graph max pooling instead of graph average pooling (ablation) | KCNet · ModelNet40 | accuracy (%): 88.0 -> 88.6 | Helped |
| add kernel correlation (ablation) | PointNet (vanilla) · ModelNet40 | accuracy (%): 87.2 -> 90.5 | Helped |
| add graph max pooling (ablation) | PointNet (vanilla) · ModelNet40 | accuracy (%): 87.2 -> 88.6 | Helped |
| add both kernel correlation and graph max pooling (ablation) | PointNet (vanilla) · ModelNet40 | accuracy (%): 87.2 -> 91.0 | Helped |
| increase the number of kernels from 16 to 32 (ablation) | KCNet · ModelNet40 | accuracy (%): 90.7 -> 91.0 | Helped |
| increase the number of kernels from 32 to 48 (ablation) | KCNet · ModelNet40 | accuracy (%): 91.0 -> 91.0 | No clear effect |
| reduce kernel points from 16 to 3 (ablation) | KCNet · ModelNet40 | accuracy (%): 91.0 -> 90.9 | No clear effect |
| reduce kernel points from 16 to 8 (ablation) | KCNet · ModelNet40 | accuracy (%): 91.0 -> 90.4 | Hurt |
| set kernel width σ to 1e-3 instead of 5e-3 (ablation) | KCNet · ModelNet40 | accuracy (%): 91.0 -> 90.0 | Hurt |
| set kernel width σ to 1e-2 instead of 5e-3 (ablation) | KCNet · ModelNet40 | accuracy (%): 91.0 -> 90.4 | Hurt |
| replace 10 input points with uniformly distributed random noise (ablation) | KCNet with kernel correlation and graph max pooling · ModelNet40 | accuracy (%): 91.0 -> 67.2 | Hurt |
Closest papers by meaning
- PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation (CVPR 2017 · similarity 0.91)
- PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space (NeurIPS 2017 · similarity 0.91)
- Convolution in the Cloud: Learning Deformable Kernels in 3D Graph Convolution Networks for Point Cloud Analysis (CVPR 2020 · similarity 0.91)
- LinNet: Linear Network for Efficient Point Cloud Representation Learning (NeurIPS 2024 · similarity 0.89)
- X-3D: Explicit 3D Structure Modeling for Point Cloud Recognition (CVPR 2024 · similarity 0.89)
- PointCFormer: A Relation-Based Progressive Feature Extraction Network for Point Cloud Completion (AAAI 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.