Research map / Image generation / Autoregressive and latent generation
Bidirectional Query-Driven Generation of Parametric CAD Sketch
Yang Liu, Daxuan Ren, Yijie Ding, Jianmin Zheng, Fang Deng
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 bidirectional query mechanism (ablation) | CADSketcher · Partial-to-full completion (SketchGraphs) | F1: 48.6 -> 59.2 | Improved |
| add hybrid positional encoding (ablation) | CADSketcher · Partial-to-full completion (SketchGraphs) | F1: 57.0 -> 59.2 | Improved |
| add confidence gate for expansion direction (ablation) | CADSketcher · Partial-to-full completion (SketchGraphs) | F1: 51.5 -> 59.2 | Improved |
| add validity compiler (ablation) | CADSketcher · Partial-to-full completion (SketchGraphs) | Invalid rate: 0.03 -> 0 | Improved |
| add validity compiler (ablation) | CADSketcher · Partial-to-full completion (SketchGraphs) | F1: 52.8 -> 59.2 | Improved |
| add iterative partial context updating (ablation) | CADSketcher · Partial-to-full completion (SketchGraphs) | F1: 50.9 -> 59.2 | Improved |
| add bidirectional query mechanism (ablation) | CADSketcher · Partial-to-full completion (SketchGraphs) | ACC_skt: 33.7 -> 45.6 | Improved |
| use CADSketcher instead of Vitruvion autoregressive baseline | parametric sketch completion · Partial-to-full completion (SketchGraphs) | ACC_tok: 35.2 -> 72.8 | Improved |
| use CADSketcher on out-of-distribution dataset | CADSketcher · Cross-dataset: CAD as a Language, partial-to-full | ACC_skt: 7.96 -> 14.1 vs Dual-AR | Improved |
| CADSketcher versus CAD-VLM on early-stage expansion coverage | CADSketcher · Early-stage sketch expansion (SketchGraphs) | COV: 75.2 -> 77.0 | Improved |
Closest papers by meaning
- CAD-Coder: Text-to-CAD Generation with Chain-of-Thought and Geometric Reward (NeurIPS 2025 · similarity 0.88)
- SketchDNN: Joint Continuous-Discrete Diffusion for CAD Sketch Generation (ICML 2025 · similarity 0.87)
- CAD-SIGNet: CAD Language Inference from Point Clouds using Layer-wise Sketch Instance Guided Attention (CVPR 2024 · similarity 0.87)
- From 2D CAD Drawings to 3D Parametric Models: A Vision-Language Approach (AAAI 2025 · similarity 0.86)
- CADMorph: Geometry‑Driven Parametric CAD Editing via a Plan–Generate–Verify Loop (NeurIPS 2025 · similarity 0.86)
- Clarify Before You Draw: Proactive Agents for Robust Text-to-CAD Generation (ICML 2026 · 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.
