Other Graph Benchmarks
ShapeNet-Part
Structure
16,881Graphs
2,048Nodes
50Classes
6Node Feature Dim
Coverage
24Results
1Variants
4Papers
- ICLR 2026: k-Maximum Inner Product Attention for Graph Transformers and the Expressive Power of GraphGPS k-Maximum Inner Product Attention for Graph Transformers and the Expressive Power of GraphGPS ICLR 2026
- Adaptive Canonicalization with Application to Invariant Anisotropic Geometric Networks Adaptive Canonicalization with Application to Invariant Anisotropic Geometric Networks
- Differentiable Graph Module (DGM) for Graph Convolutional Networks Differentiable Graph Module (DGM) for Graph Convolutional Networks
- Dynamic Graph CNN for Learning on Point Clouds Dynamic Graph CNN for Learning on Point Clouds
50-class node classification on 3D point clouds, predicting a part label for every point across 16 object categories and 50 part classes. Evaluated by F1.
Dataset details
Task
3D point cloud part segmentation on ShapeNet objects. Graph-based part label prediction on point cloud graphs (16 object categories, 50 part classes). The primary catalog metric is F1.
Data
3D point cloud part segmentation on ShapeNet objects. Graph-based part label prediction on point cloud graphs (16 object categories, 50 part classes). The cataloged artifact reports 50 target classes or tasks.
Features
Input construction is dataset-specific and follows the cited release: program call graphs, 3D point coordinates, text-derived vectors, product attributes, or movie metadata. Added edges, neighborhood graphs, and feature normalization are preprocessing choices.
Splits and evaluation
The catalog records these protocols or variants: Standard split. Evaluation uses F1 (higher is better). Exact masks or folds must come from the cited release.
Quirks and pitfalls
These entries come from unrelated upstream benchmarks and have no shared protocol. Report the exact graph construction, split, feature pipeline, and label granularity; similarly named Film, Actor, IMDB, or Amazon artifacts are not interchangeable.
Sources
SOTA Progression
new SOTA
matches SOTA at that time
other proposed result
other baseline result
Leaderboard
| # | Model | F1 ↑ | mIoU | Paper |
|---|---|---|---|---|
| 1 | GPS + k-MIP | 0.8268
±
0.0064 | — | |
| 2 | Exphormer | 0.8262
±
0.0031 | — | |
| 3 | VN-DGCNN | 0.8140 | — | |
| 4 | AC-DGCNN | 0.8090
±
0.0070 | — | |
| 5 | GPS + BigBird | 0.7965
±
0.0098 | — | |
| 6 | DGCNN | 0.7860 | — | |
| 7 | CN-DGCNN | 0.7850
±
0.0090 | — | |
| 8 | GPS + Performer | 0.7736
±
0.0123 | — | |
| 9 | GatedGCN | 0.7620
±
0.0032 | — | |
| 10 | AC-PointNet | 0.7600
±
0.0060 | — | |
| 11 | CN-PointNet | 0.7360
±
0.0110 | — | |
| 12 | VN-PointNet | 0.7280 | — | |
| 13 | GINE | 0.6457
±
0.0035 | — | |
| 14 | GAT | 0.6301
±
0.0017 | — | |
| 15 | PointNet++ | 0.6230 | — | |
| 16 | GCN | 0.6018
±
0.0004 | — | |
| 17 | DGCNN | — mIoU
0.8520 | 0.8520 | |
| 18 | dDGM | — mIoU
0.8560 | 0.8560 | |
| 19 | PointNet++ | — mIoU
0.8510 | 0.8510 | |
| 20 | Kd-Net | — mIoU
0.8230 | 0.8230 | |
| 21 | LocalFeatureNet | — mIoU
0.8430 | 0.8430 | |
| 22 | PCNN | — mIoU
0.8510 | 0.8510 | |
| 23 | PointCNN | — mIoU
0.8610 | 0.8610 | |
| 24 | DGCNN | — mIoU
0.8520 | 0.8520 |
SOTA performance
set a new SOTA at the time
matched the SOTA at the time