V2G pipeline converts CAD diagrams into property graphs, capturing component topology and connectivity overlooked by pixel-based MLLMs. It delivers major accuracy gains on electrical schematic compliance benchmarks where top MLLMs fail. Benchmark and code released on GitHub for further research.
Key Points
- 1.Vector-to-graph conversion for schematics
- 2.Outperforms MLLMs on compliance auditing
- 3.Open-sources benchmark and implementation
Impact Analysis
Addresses structural blindness in multimodal AI, enabling reliable engineering audits. Paves way for practical AI deployment in CAD and schematic analysis domains.
Technical Details
Nodes represent components, edges encode connectivity in property graphs. Tested on diagnostic benchmark for electrical error categories. arXiv:2602.11678v1.