Research map / Point clouds / Point cloud classification
Rotation-Invariant Local-to-Global Representation Learning for 3D Point Cloud
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 |
|---|---|---|---|
| Use rotation-invariant local descriptors and hierarchical GCN aggregation | RI-GCN · ModelNet40 z/SO(3) | accuracy: RI-GCN 89.5; RIConv 86.4 | Helped |
| Use RI-GCN for part segmentation | ShapeNet · ShapeNet z/SO(3) | part segmentation performance: 77.2 vs. RIConv 75.3 | Helped |
| Make dilation rate, neighbor count, and graph edge count stochastic (ablation) | RI-GCN · ModelNet40 z/SO(3) | accuracy: 87.7 -> 89.5 | Helped |
| Replace GCN layers with fully connected layers (ablation) | RI-GCN · ModelNet40 z/SO(3) | accuracy: 89.1 vs. 89.5 | Hurt |
| Replace local reference frames with a global PCA rotation (ablation) | RI-GCN · ModelNet40 z/SO(3) | accuracy: 87.2 vs. 89.5 | Hurt |
| Use two hierarchy levels instead of three (ablation) | RI-GCN · ModelNet40 z/SO(3) | accuracy: 89.1 vs. 89.5 | Hurt |
| Use four hierarchy levels instead of three (ablation) | RI-GCN · ModelNet40 z/SO(3) | accuracy: 89.3 vs. 89.5 | Hurt |
| Replace stochastic dilated k-NN with deterministic dilated k-NN (ablation) | RI-GCN · ModelNet40 z/SO(3) | accuracy: 88.3 vs. 89.5 | Hurt |
| Use stochastic dilation rate only (ablation) | RI-GCN · ModelNet40 z/SO(3) | accuracy: 87.7 -> 88.6 | Helped |
| Use stochastic neighbor count only (ablation) | RI-GCN · ModelNet40 z/SO(3) | accuracy: 87.7 -> 88.9 | Helped |
| Use stochastic graph edge count only (ablation) | RI-GCN · ModelNet40 z/SO(3) | accuracy: 87.7 -> 88.0 | Helped |
Closest papers by meaning
- ClusterNet: Deep Hierarchical Cluster Network With Rigorously Rotation-Invariant Representation for Point Cloud Analysis (CVPR 2019 · similarity 0.91)
- Rethinking Rotation Invariance with Point Cloud Registration (AAAI 2023 · similarity 0.89)
- Convolution in the Cloud: Learning Deformable Kernels in 3D Graph Convolution Networks for Point Cloud Analysis (CVPR 2020 · similarity 0.89)
- Local-consistent Transformation Learning for Rotation-invariant Point Cloud Analysis (CVPR 2024 · similarity 0.87)
- TetraSphere: A Neural Descriptor for O(3)-Invariant Point Cloud Analysis (CVPR 2024 · similarity 0.87)
- RI-MAE: Rotation-Invariant Masked AutoEncoders for Self-Supervised Point Cloud Representation Learning (AAAI 2025 · similarity 0.86)
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.