VLM Revolutionizes Building Layout Representation with Multi-Granularity Graph Extraction
Summary:This research introduces a Vision-Language Model (VLM)-based approach for automatic multi-granularity graph representation of building layouts. The method employs node identification, edge inference, text parsing, and graph coarsening to extract structured information from architectural floor plans. Experiments demonstrate that VLM-generated graphs are highly consistent with human-labeled graphs and perform well in real-world tasks. This scalable, annotation-free method offers a novel solution f
Background and Challenges
Architectural floor plans encode rich relational knowledge among functional spaces, which is crucial for design retrieval, knowledge-based reasoning, and BIM enrichment. However, automatically constructing task-adaptive graph representations for public buildings remains a challenge.
Methodology and Innovation
The study presents a VLM-based approach for automatic multi-granularity graph representation, which includes the following steps:
- Node Identification: VLM identifies functional space nodes from floor plans.
- Edge Inference: VLM infers relationships between nodes to generate graph edges.
- Text Parsing: Parses textual information in floor plans to extract key semantic details.
- Graph Coarsening: Coarsens the graph based on granularity requirements to produce multi-granularity representations.
Experiments and Results
Using 147 academic library floor plans worldwide as a case study, the results show that:
- VLM-generated graphs are highly consistent with human-labeled graphs, with a matched node ratio exceeding 92%.
- The average time for generating a three-granularity graph per floor plan is 509.3 seconds.
- Meso-grained graphs yield the best node-level zone prediction (Macro F1 = 0.647, at 65% of fine-grained complexity).
- Coarse-grained graphs are most effective for graph-level layout quality evaluation (Spearman's ρ = 0.610, at 16% of fine-grained complexity).
Industry Impact and Future Directions
This research converts floor plan images into knowledge representations, enabling scalable, annotation-free extraction of structured layout information. It significantly enhances the efficiency of design informatics and is expected to play a key role in areas such as architectural design, smart city planning, and building asset management.
Developer Recommendations
- Expand Application Scenarios: Developers can apply this method to other types of architectural drawings (e.g., factory layouts, mall designs) to further validate its versatility.
- Model Optimization: Optimize the VLM model to improve the accuracy and efficiency of graph generation based on specific domain needs.
- Cross-Platform Integration: Integrate this method with existing BIM platforms to enable seamless transfer and sharing of design information.
— END —Source: ArXiv AI (cs.AI) (2026-08-27)
Tags: #VLM #Architectural Design #Graph Generation #AI Agents #Design Informatics
Editorial & Fact-Checking Note: This article is compiled from primary research, official release documentation, and source papers by the ZICQ Newsroom pipeline with automated entity verification and human editorial review. If you notice any technical inaccuracy, please submit a correction via our corrections policy or email our editorial desk directly.
Community Comments