VFIG: Vectorizing Complex Figures in SVG with Vision-Language Models

Qijia He, Xunmei Liu, Hammaad Memon, Ziang Li, Zixian Ma et al.|March 25, 2026arXiv

Key Takeaway

You can now automatically convert flat images of technical figures into editable, scalable vector graphics—matching GPT-5.2 performance—enabling recovery of lost design source files without manual reconstruction.

Summary

VFIG converts rasterized images (PNG, JPEG) of technical diagrams back into editable SVG vector graphics using vision-language models. The team created a 66K dataset of figure-SVG pairs and a two-stage training approach (supervised learning for basic shapes, then reinforcement learning for refinement) to reconstruct complex professional diagrams with high fidelity.

multimodal training applications

Key Terms

vision-language-model vector-graphics coarse-to-fine-training reinforcement-learning curated-dataset