GVR-Coder: A Visual-Feedback Framework for Structured SVG Generation in Complex Document and Meeting Scenarios

2026-07-30

AI Summary

This work addresses three key challenges in generating structured SVG diagrams from long-form text in complex document and meeting scenarios: scarcity of logical diagram data, spatial disarray due to missing layout priors, and the absence of fine-grained visual feedback for correcting aesthetic flaws. To overcome these issues, the authors propose GVR-Coder, a novel framework that integrates explicit layout constraints, curriculum-driven rejection sampling fine-tuning, dual-rendering feedback reinforcement learning, and a Generate-Validate-Repair (GVR) agent loop to establish a new paradigm for high-quality SVG generation. Evaluated on the newly curated DocMeetSVG-100K large-scale dataset, the method significantly outperforms existing baselines in both logical coherence and visual aesthetics. The code and dataset are publicly released.

Abstract

In demanding professional environments and meeting review scenarios, lengthy text often imposes a high cognitive load. To facilitate efficient information communication, transforming verbose text into logically clear diagrams is essential. Scalable Vector Graphics (SVG) provide an effective representation for this purpose due to their editability and resolution independence. However, current research on Text-to-SVG generation remains hindered by three major challenges: (1) the scarcity of datasets for complex, logic-rich diagrams; (2) the absence of explicit layout priors, which leads to chaotic spatial arrangements; and (3) the lack of fine-grained visual feedback to validate rendered outputs and correct aesthetic defects. To address these challenges, at the data level, we introduce DocMeetSVG-100K, a large-scale SVG dataset tailored for document authoring and meeting review scenarios. At the model level, we propose GVR-Coder, a novel framework designed to generate high-quality logical diagrams from lengthy professional texts. Specifically, we adopt a curriculum-driven rejection sampling fine-tuning to progressively enhance the model's capability in modeling complex structures, while explicitly incorporating layout constraint knowledge during training. In addition, we introduce reinforcement learning from dual rendering feedback, a mechanism that provides implicit feedback through reward signals to jointly optimize structural complexity and visual aesthetics. Furthermore, we design a generate-verify-repair agent loop, which improves generation quality through explicit, fine-grained feedback and targeted refinement. Extensive experiments demonstrate that GVR-Coder outperforms competitive baselines and reliably produces logically coherent and visually appealing diagrams. Code and data are available at https://github.com/CurryaNa/GVR-Coder.

Problem

• Text-to-SVG generation
• complex diagrams
• layout priors
• visual feedback
• SVG dataset

Innovation

• Text-to-SVG
• layout constraints
• reinforcement learning
• visual feedback
• curriculum-driven fine-tuning