Chart2SVG: Editable SVG Generation from Raster Chart Images

Authors

Jinning Cui (Renmin University of China), Lu Chen (Zhejiang University), Haoyan Shi (Shandong University), Yue He (Renmin University), Chenglong Wang (Microsoft Research), Mengyu Zhou (Alibaba), Weidong Huang (University of Technology Sydney), Yunhai Wang (Renmin University of China)

Presentation

Session: How can I do this myself?
Time: Thursday, Nov 12, 09:12 – 09:24 (US/Eastern) · session 08:00 – 09:30
Location: Hall America center

Keywords

Chart Reverse Engineering, Scalable Vector Graphics, Vision-Language Models

Abstract

We present Chart2SVG, a multimodal large language model that converts static raster charts into structurally organized, semantically enriched SVGs that support programmatic editing. By incorporating chart-specific semantic tokens into a vision-language model, Chart2SVG captures both geometric primitives and their functional roles. To support robust structural recovery, we introduce Beagle+, a dataset of 33K canonicalized and structurally distilled chart samples. Our approach combines specialized training objectives with a rendering-aware post-training phase, producing SVGs that are both visually accurate and structurally consistent. To facilitate higher-level manipulations, we construct a Chart Structure Graph (CSG) that exposes visual dependencies, enabling tasks such as interactive exploration, chart repurposing, and layout reuse. Experiments show that Chart2SVG substantially outperforms baselines in reconstruction fidelity and downstream editing utility, advancing the development of intelligent and interactive visualization tools.

For Practitioners

data scientists, researchers, chart disigners