Research map / Vision-language and multimodal / Multimodal LLMs
OmniSch: A Multimodal PCB Schematic Benchmark For Structured Diagram Visual Reasoning
Taiting Lu, Kaiyuan Lin, Yuxin Tian, Yubo Wang, Muchuan Wang, Sharique Khatri, Akshit Kartik, Yixi Wang, Amey Rane, Yida Wang, Yifan Yang, Yi-Chao Chen and 2 more
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| keep text labels on the schematic (ablation) | Claude Sonnet 4.6 · synthetic schematic symbol detection | F1: 0.7235 -> 0.6247 | Hurt |
| keep text labels on the schematic (ablation) | Gemini 2.5 Flash · synthetic schematic symbol detection | F1: 0.7241 -> 0.5622 | Hurt |
| add few-shot example schematics (ablation) | GPT-5.2 · synthetic schematic symbol detection | F1: 0.4732 -> 0.6196 | Improved |
| add few-shot example schematics (ablation) | Claude Sonnet 4.6 · synthetic schematic symbol detection | F1: 0.6247 -> 0.7185 | Improved |
| add few-shot example schematics (ablation) | Gemini 2.5 Flash · synthetic schematic symbol detection | F1: 0.5622 -> 0.7469 | Improved |
| add few-shot example schematics (ablation) | Gemini 2.5 Flash · synthetic schematic geometry | Kendall tau: 0.0219 -> -0.0965 | Hurt |
| use full resolution instead of low resolution (ablation) | GPT-5.2 · real schematic symbol detection | F1: 0.3184 -> 0.5276 | Improved |
| use full resolution instead of low resolution (ablation) | Claude Sonnet 4.6 · real schematic symbol detection | F1: 0.5130 -> 0.5862 | Improved |
| provide ground-truth bounding box to the agent (ablation) | GPT-5.2 · agentic symbol name | accuracy: 0.6948 -> 0.8565 | Improved |
| provide ground-truth bounding box to the agent (ablation) | GPT-5 Mini · agentic symbol name | accuracy: 0.5539 -> 0.8932 | Improved |
| provide ground-truth bounding box to the agent (ablation) | Gemini-3.1-Pro-Preview · agentic symbol name | accuracy: 0.8093 -> 0.8958 | Improved |
| provide ground-truth bounding box to the agent (ablation) | Claude-Sonnet-4.6 · agentic symbol type | accuracy: 0.7452 -> 0.7202 | Hurt |
Closest papers by meaning
- CircuitSense: A Hierarchical Circuit System Benchmark Bridging Visual Comprehension and Symbolic Reasoning in Engineering Design Process (ICLR 2026 · similarity 0.89)
- VisionGraph: Leveraging Large Multimodal Models for Graph Theory Problems in Visual Context (ICML 2024 · similarity 0.87)
- MAPS: Advancing Multi-Modal Reasoning in Expert-Level Physical Science (ICLR 2025 · similarity 0.86)
- VCG-Bench: Towards A Unified Visual-Centric Benchmark for Structured Generation and Editing (ICML 2026 · similarity 0.85)
- VisioMath: Benchmarking Figure-based Mathematical Reasoning in LMMs (ICLR 2026 · similarity 0.85)
- Functional Matching of Logic Subgraphs: Beyond Structural Isomorphism (NeurIPS 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
