Research map / Image generation / Autoregressive and latent generation
UniSketch: A Unified Framework for Parametric Sketch Generation and Constraint Prediction
Jing Lin, Fazhi He, Rubin Fan
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 |
|---|---|---|---|
| use UniSketch instead of Vitruvion primitive network | image-to-sketch generation · Vitruvion dataset (original images) | primitive accuracy (TF): 43.97 -> 51.95 | Improved |
| use UniSketch instead of Vitruvion constraint network | constraint prediction · Vitruvion dataset | constraint accuracy (TF): 74.69 -> 79.02 | Improved |
| use UniSketch instead of SketchGraphs | constraint prediction · Vitruvion dataset | F1 (AR): 52.16 -> 66.00 | Improved |
| mixed-batch training (p=0.5 image) instead of image-sequence only (ablation) | UniSketch · UniSketch 1M subset | primitive accuracy: 29.71 -> 36.61 | Improved |
| mixed-batch training instead of image-sequence only (ablation) | UniSketch · UniSketch 1M subset | constraint accuracy: 64.94 vs 64.84 | No clear effect |
| mixed-batch instead of sequence-only training (ablation) | UniSketch · UniSketch 1M subset | primitive accuracy: 29.97 -> 36.61 | Improved |
| add dual output heads (primitive and constraint) (ablation) | UniSketch · UniSketch 1M subset | primitive accuracy: 29.58 -> 29.97 | Improved |
| add dual output heads (ablation) | UniSketch · UniSketch 1M subset | constraint accuracy: 64.72 -> 64.80 | Improved |
Closest papers by meaning
- MiCADangelo: Fine-Grained Reconstruction of Constrained CAD Models from 3D Scans (NeurIPS 2025 · similarity 0.88)
- CADCrafter: Generating Computer-Aided Design Models from Unconstrained Images (CVPR 2025 · similarity 0.87)
- From 2D CAD Drawings to 3D Parametric Models: A Vision-Language Approach (AAAI 2025 · similarity 0.87)
- Revisiting CAD Model Generation by Learning Raster Sketch (AAAI 2025 · similarity 0.87)
- 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 0.86)
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.
