Research map / Point clouds / Point cloud classification
Bridging Tokens and Geometry: Token-wise 3D Supervision for CAD Generation
Yijia Guan, Jianhua Sun
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 A3PL token-wise 3D point loss (ablation) | Text2CAD · DeepCAD dataset | CD mean: 30.5 -> 25.2 | Improved |
| add A3PL token-wise 3D point loss (ablation) | DeepCAD · DeepCAD dataset | CD mean: 7.1 -> 8.9 | Hurt |
| add extrusion sequence augmentation (ESAug) on top of A3PL (ablation) | Text2CAD · DeepCAD dataset | CD mean: 25.2 -> 23.9 | Improved |
| add ESAug on top of A3PL (ablation) | DeepCAD · DeepCAD dataset | CD mean: 8.9 -> 7.3 | Improved |
| add ESAug on top of A3PL (ablation) | Drawing2CAD · DeepCAD dataset | IR: 2.6 -> 2.2 | Improved |
| add grammar-constrained operator (GCO) alone (ablation) | Text2CAD · DeepCAD dataset | IR: 3.5 -> 2.2 | Improved |
| add GCO alone (ablation) | CADFusion · DeepCAD dataset | IR: 6.0 -> 2.5 | Improved |
| add GCO alone (ablation) | CADFusion · DeepCAD dataset | CD mean: 20.8 -> 21.4 | Hurt |
| add GCO alone (ablation) | Drawing2CAD · DeepCAD dataset | IR: 20. -> 1.9 | Improved |
| combine A3PL, ESAug and GCO versus A3PL+ESAug (ablation) | Text2CAD · DeepCAD dataset | F1 Sketch: 70.2 -> 74.8 | Improved |
| combine A3PL, ESAug and GCO versus baseline (ablation) | CADFusion · DeepCAD dataset | IR: 6.0 -> 1.9 | Improved |
| combine A3PL, ESAug and GCO versus baseline (ablation) | DeepCAD · DeepCAD dataset | CD mean: 7.1 -> 6.8 | Improved |
Closest papers by meaning
- CAD-Coder: Text-to-CAD Generation with Chain-of-Thought and Geometric Reward (NeurIPS 2025 · similarity 0.88)
- GeoCAD: Local Geometry-Controllable CAD Generation with Large Language Models (NeurIPS 2025 · similarity 0.88)
- CAD-Llama: Leveraging Large Language Models for Computer-Aided Design Parametric 3D Model Generation (CVPR 2025 · similarity 0.88)
- Generating CAD Code with Vision-Language Models for 3D Designs (ICLR 2025 · similarity 0.87)
- FlexCAD: Unified and Versatile Controllable CAD Generation with Fine-tuned Large Language Models (ICLR 2025 · similarity 0.87)
- StepCAD: Mesh-to-CAD Code Generation via LLM Policy and Geometry-Guided Search (NeurIPS 2026 · similarity 0.87)
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.
