Research map / Vision-language and multimodal / CLIP and zero-shot
Semantic Document Derendering: SVG Reconstruction via Vision-Language Modeling
Adam Hazimeh, Ke Wang, Mark Collier, Gilles Baechler, Efi Kokiopoulou, Pascal Frossard
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 YOLO bounding box priors (ablation) | SliDer Gemini · Slide2SVG test (225 slides) | mIoU (%): 81.90 -> 89.71 | Improved |
| add YOLO bounding box priors (ablation) | SliDer Gemini · Slide2SVG test (225 slides) | OCR accuracy (%): 89.78 -> 92.42 | Improved |
| add YOLO bounding box priors (ablation) | SliDer Gemini · Slide2SVG test (225 slides) | LPIPS: 0.090 -> 0.072 | Improved |
| add YOLO bounding box priors (ablation) | SliDer Gemma · Slide2SVG test (225 slides) | mIoU (%): 69.64 -> 89.04 | Improved |
| add YOLO bounding box priors (ablation) | SliDer Gemma · Slide2SVG test (225 slides) | LPIPS: 0.105 -> 0.076 | Improved |
| add one step of iterative refinement (with YOLO priors) (ablation) | SliDer Gemini · Slide2SVG test (225 slides) | LPIPS: 0.072 -> 0.069 | Improved |
| add one step of iterative refinement (with YOLO priors) (ablation) | SliDer Gemini · Slide2SVG test (225 slides) | mIoU (%): 89.71 -> 89.14 | Hurt |
| add one step of iterative refinement (no YOLO priors) (ablation) | SliDer Gemini · Slide2SVG test (225 slides) | OCR accuracy (%): 89.78 -> 90.57 | Improved |
| add one step of iterative refinement (no YOLO priors) (ablation) | SliDer Gemini · Slide2SVG test (225 slides) | LPIPS: 0.090 -> 0.088 | Improved |
| add one step of iterative refinement (with YOLO priors) (ablation) | SliDer Gemma · Slide2SVG test (225 slides) | OCR accuracy (%): 92.90 -> 93.53 | Improved |
| add one step of iterative refinement (with YOLO priors) (ablation) | SliDer Gemma · Slide2SVG test (225 slides) | MSE: 13.34 -> 13.38 | No clear effect |
| add one step of iterative refinement (no YOLO priors) (ablation) | SliDer Gemma · Slide2SVG test (225 slides) | LPIPS: 0.105 -> 0.105 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- LIVE — LIVE: Learnable In-Context Vector for Visual Question Answering (NeurIPS 2024) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- VFIG: Vectorizing Complex Figures in SVG with Vision-Language Models (NeurIPS 2026 · similarity 0.86)
- 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 0.86)
- Vector Grimoire: Codebook-based Shape Generation under Raster Image Supervision (ICML 2025 · similarity 0.85)
- InternSVG: Towards Unified SVG Tasks with Multimodal Large Language Models (ICLR 2026 · similarity 0.85)
- VisMin: Visual Minimal-Change Understanding (NeurIPS 2024 · 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.
