Research map / Point clouds / Point cloud classification
BRepNet: A Topological Message Passing System for Solid Models
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 |
|---|---|---|---|
| Use the asymmetric kernel with loop-order information rather than the simple edge kernel (ablation) | BRepNet · Fusion 360 Gallery | IoU: 0.98% improvement in IoU | Helped |
| Use the winged edge kernel rather than the simple edge kernel (ablation) | BRepNet · Fusion 360 Gallery | IoU: over 5 standard deviations above the IoU value achieved by the “simple edge” kernel | Helped |
| Expand the kernel beyond the winged edge configuration (ablation) | BRepNet · Fusion 360 Gallery | IoU | No clear effect |
| Remove surface type one-hot encodings (ablation) | BRepNet · Fusion 360 Gallery | IoU: 3.7% reduction in IoU | Hurt |
| Remove curve type information (ablation) | BRepNet · Fusion 360 Gallery | IoU: 3.9% reduction in IoU | Hurt |
| Remove edge convexity (ablation) | BRepNet · Fusion 360 Gallery | IoU: 4.6% reduction in IoU | Hurt |
| Remove edge length (ablation) | BRepNet · Fusion 360 Gallery | IoU: 0.7% decrease | Hurt |
| Remove face area (ablation) | BRepNet · Fusion 360 Gallery | IoU: 0.4% IoU decrease | Hurt |
| Use BRepNet's coedge-relative topological convolution instead of ECC | BRepNet versus ECC · Fusion 360 Gallery | per-face IoU: 77.10% vs 72.08% | Helped |
| Use direct B-rep input rather than point-cloud input | BRepNet versus PointNet++ · Fusion 360 Gallery | per-face IoU: 77.10 ± 0.54 vs 34.78 ± 2.23 | Helped |
| Use direct B-rep input rather than mesh input | BRepNet versus MeshCNN · Fusion 360 Gallery | per-face IoU: 77.10 ± 0.54 vs 20.59 ± 0.41 | Helped |
| Pool coedge information onto parent faces and edges between convolution units (ablation) | BRepNet · Fusion 360 Gallery | performance | Helped |
Closest papers by meaning
- Brep2Shape: Boundary and Shape Representation Alignment via Self-supervised Transformers (ICML 2026 · similarity 0.92)
- BrepGiff: Lightweight Generation of Complex B-rep with 3D GAT Diffusion (CVPR 2025 · similarity 0.90)
- UV-Net: Learning From Boundary Representations (CVPR 2021 · similarity 0.90)
- DTGBrepGen: A Novel B-rep Generative Model through Decoupling Topology and Geometry (CVPR 2025 · similarity 0.88)
- SpelsNet: Surface Primitive Elements Segmentation by B-Rep Graph Structure Supervision (NeurIPS 2024 · similarity 0.87)
- SegGraph: Leveraging Graphs of SAM Segments for Few-Shot 3D Part Segmentation (NeurIPS 2025 · similarity 0.83)
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.