Research map / Point clouds / Point cloud classification
UV-Net: Learning From Boundary Representations
Check 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 |
|---|---|---|---|
| represent B-rep geometry as UV-grids and topology as an adjacency graph, then jointly process them | UV-Net · Machining Feature classification | Accuracy (%): 87.13 $\pm$ 0.15 -> 99.94 $\pm$ 0.00 | Helped |
| represent B-rep geometry as UV-grids and topology as an adjacency graph, then jointly process them | UV-Net · FabWave classification | Accuracy (%): 80.08 $\pm$ 3.61 -> 94.51 $\pm$ 0.10 | Helped |
| represent B-rep geometry as UV-grids and topology as an adjacency graph, then jointly process them | UV-Net · SolidLetters classification | Accuracy (%): 96.62 $\pm$ 0.13 -> 97.24 $\pm$ 0.10 | Helped |
| segment directly from B-rep faces using UV-Net | MFCAD · MFCAD | Per-face IoU: 99.70 $\pm$ 0.06 -> 99.87 $\pm$ 0.03 | Helped |
| segment directly from B-rep faces using UV-Net | ABC · ABC | Per-face IoU: 18.14 $\pm$ 1.97 -> 50.37 $\pm$ 1.11 | Helped |
| add surface normals to the input features (ablation) | UV-Net segmentation · MFCAD | Per-face IoU: 99.63 $\pm$ 0.06 -> 99.87 $\pm$ 0.03 | Helped |
| add surface normals to the input features (ablation) | UV-Net segmentation · ABC | Per-face IoU: 38.99 $\pm$ 0.42 -> 50.37 $\pm$ 1.11 | Helped |
| use D2-equivariant convolutions and group pooling instead of regular convolutions (ablation) | UV-Net surface CNN · SolidLetters test set with reparametrized surfaces | Test accuracy: 55.98 $\pm$ 2.36 -> 96.59 $\pm$ 0.02 | Helped |
| apply UV flips and rotations to test surfaces without training on them (ablation) | UV-Net with regular convolutions · SolidLetters test set | Test accuracy: 96.74 $\pm$ 0.06 -> 55.98 $\pm$ 2.36 | Hurt |
| use D2-equivariant convolutions under reparametrization (ablation) | UV-Net surface CNN · SolidLetters test set | Test accuracy: 96.58 $\pm$ 0.01 -> 96.59 $\pm$ 0.02 | No clear effect |
| remove curve features, curve CNN, and edge-conditioned message passing (ablation) | UV-Net segmentation · MFCAD | IoU | Hurt |
| replace graph message passing with an MLP and remove edge features (ablation) | UV-Net segmentation · MFCAD | IoU | Hurt |
Closest papers by meaning
- DTGBrepGen: A Novel B-rep Generative Model through Decoupling Topology and Geometry (CVPR 2025 · similarity 0.87)
- SpelsNet: Surface Primitive Elements Segmentation by B-Rep Graph Structure Supervision (NeurIPS 2024 · similarity 0.87)
- PolyhedronNet: Representation Learning for Polyhedra with Surface-attributed Graph (ICLR 2025 · similarity 0.86)
- Geometry-Informed Neural Networks (ICML 2025 · similarity 0.86)
- BrepGiff: Lightweight Generation of Complex B-rep with 3D GAT Diffusion (CVPR 2025 · similarity 0.86)
- 3D Neural Edge Reconstruction (CVPR 2024 · 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.