Compositional SVG Generation via VLM-Driven Hierarchical Semantic Parsing
On the map: LLM reasoning → Multimodal reasoning
Abstract (arXiv)
While Vision-Language Models (VLMs) excel at visual reasoning, generating structured, editable Scalable Vector Graphics (SVG) remains a fundamental challenge. Existing pipelines predominantly yield flat, semantically agnostic collections of paths, where editing a single object requires manually identifying its constituent paths. To address this, we propose a VLM-driven agentic framework for semantic compositional SVG generation. Our pipeline recursively parses visual scenes into semantic and geometric hierarchies via top-down decomposition, visual grounding, and prompt-driven amodal occlusion recovery, ensuring each component is geometrically complete. Furthermore, we introduce the Semantic SVG Benchmark with human-annotated semantic groups and novel sub-component metrics (Semantic Recall/Precision, PERE) to explicitly evaluate structural compositionality and functional editability. Experiments show that our natively predicted structures surpass the upper bounds of existing flat-generation methods in both grouping quality and editability, while maintaining state-of-the-art visual fidelity.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Multimodal reasoning, one of 17 directions in LLM reasoning. Builds on 19 of its 33 references that are in the map. First posted on arXiv 13 Sep 2026.
Builds on
19 of 33 references.
Closest ancestors · 1
Same direction, newest first
Other references · 18
- 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,695
- 2023Toolformer: Language Models Can Teach Themselves to Use Toolscited by 4,182
- 2024Scaling Rectified Flow Transformers for High-Resolution Image Synthesiscited by 4,181
- 2023ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIscited by 1,947
- 2025InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Modelscited by 1,834
- 2025SAM 3: Segment Anything with Conceptscited by 906
- 2024Executable Code Actions Elicit Better LLM Agentscited by 564
- 2023LLaVA-Plus: Learning to Use Tools for Creating Multimodal Agentscited by 193
- 2023StarVector: Generating Scalable Vector Graphics Code from Images and Textcited by 102
- 2025OmniSVG: A Unified Scalable Vector Graphics Generation Modelcited by 74
- 2024Empowering LLMs to Understand and Generate Complex Vector Graphicscited by 66
- 2025Rendering-Aware Reinforcement Learning for Vector Graphics Generationcited by 39
6 more
- 2025AgentMath: Empowering Mathematical Reasoning for Large Language Models via Tool-Augmented Agentcited by 17
- 2026The Evolution of Tool Use in LLM Agents: From Single-Tool Call to Multi-Tool Orchestrationcited by 14
- 2025LayerPeeler: Autoregressive Peeling for Layer-wise Image Vectorizationcited by 8
- 2025DuetSVG: Unified Multimodal SVG Generation with Internal Visual Guidancecited by 8
- 2026VectorGym: A Multi-Task Benchmark for SVG Code Generation, Sketching and Editingcited by 5
- 2025Vector Prism: Animating Vector Graphics by Stratifying Semantic Structurecited by 2
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2026Evaluating Constrained Iterative Refinement for Scalable Vector Graphics Generation with Off-the-Shelf VLMs
- 2025SVGThinker: Instruction-Aligned and Reasoning-Driven Text-to-SVG Generation
- 2024Chat2SVG: Vector Graphics Generation with Large Language Models and Image Diffusion Models
- 2023Beyond Pixels: Exploring Human-Readable SVG Generation for Simple Images with Vision Language Models
- 2026SVG-Score: Human-Aligned Evaluation of Text-to-SVG Generation
- 2025UniSVG: A Unified Dataset for Vector Graphic Understanding and Generation with Multimodal Large Language Models
- 2025VectorEdits: A Dataset and Benchmark for Instruction-Based Editing of Vector Graphics
- 2025SVGen: Interpretable Vector Graphics Generation with Large Language Models