Research map / Image generation / Autoregressive and latent generation
CanvasVAE: Learning To Generate Vector Graphic Documents
Kota Yamaguchi
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 one-shot decoding instead of autoregressive decoding (Transformer x1) (ablation) | CanvasVAE · Crello test | S_reconst: 85.47 -> 88.67 | Improved |
| use one-shot decoding instead of autoregressive decoding (Transformer x1) (ablation) | CanvasVAE · Crello test | mIoU: 33.86 -> 47.02 | Improved |
| use one-shot decoding instead of autoregressive decoding (LSTM) (ablation) | LSTM VAE · Crello test | mIoU: 33.52 -> 40.50 | Improved |
| use one-shot decoding instead of autoregressive decoding (Transformer x1) (ablation) | CanvasVAE · RICO test | mIoU: 51.13 -> 60.42 | Improved |
| use one-shot decoding instead of autoregressive decoding (Transformer x1) (ablation) | CanvasVAE · RICO test | S_reconst: 94.96 -> 94.35 | Hurt |
| use Transformer blocks instead of LSTM (one-shot) (ablation) | CanvasVAE · Crello test | S_reconst: 84.95 -> 88.67 | Improved |
| use Transformer blocks instead of LSTM (one-shot) (ablation) | CanvasVAE · RICO test | S_gen: 92.05 -> 93.90 | Improved |
| use Transformer blocks instead of LSTM (autoregressive) (ablation) | AutoReg model · Crello test | S_reconst: 79.75 -> 85.47 | Improved |
| stack 4 Transformer blocks instead of 1 (one-shot) (ablation) | CanvasVAE · Crello test | S_gen: 87.57 -> 88.15 | Improved |
| stack 4 Transformer blocks instead of 1 (one-shot) (ablation) | CanvasVAE · Crello test | S_reconst: 88.67 -> 87.75 | Hurt |
| stack 4 Transformer blocks instead of 1 (one-shot) (ablation) | CanvasVAE · RICO test | S_gen: 93.90 -> 95.14 | Improved |
| stack 4 Transformer blocks instead of 1 (one-shot) (ablation) | CanvasVAE · RICO test | mIoU: 60.42 -> 59.47 | Hurt |
Closest papers by meaning
- Vector Grimoire: Codebook-based Shape Generation under Raster Image Supervision (ICML 2025 · similarity 0.88)
- Chat2SVG: Vector Graphics Generation with Large Language Models and Image Diffusion Models (CVPR 2025 · similarity 0.86)
- SVGDreamer: Text Guided SVG Generation with Diffusion Model (CVPR 2024 · similarity 0.85)
- VecGlypher: Unified Vector Glyph Generation with Language Models (CVPR 2026 · similarity 0.84)
- AutomaTikZ: Text-Guided Synthesis of Scientific Vector Graphics with TikZ (ICLR 2024 · similarity 0.84)
- DesignDiffusion: High-Quality Text-to-Design Image Generation with Diffusion Models (CVPR 2025 · similarity 0.84)
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.
