AMSnet 2.0: A Large AMS Database with AI Segmentation for Net Detection

2025-05-14

AI Summary

Current multimodal large language models (MLLMs) struggle to comprehend analog/mixed-signal (AMS) circuit schematics due to the scarcity of high-quality schematic-netlist paired data; conventional approaches (e.g., AMSnet) rely on hand-crafted rules, exhibiting poor robustness and limited generalizability. This work proposes the first AI-driven AMS netlist extraction framework. It introduces a novel semantic segmentation–based wire detection mechanism to achieve robust topological identification and precise spatial coordinate recovery. We construct AMSnet 2.0—a large-scale, multi-modal dataset comprising 2,686 schematics (+239% over AMSnet), uniquely annotated with schematic images, Spectre netlists, OpenAccess layout files, and component- and net-level spatial coordinates. The framework enables end-to-end digital schematic reconstruction. Experiments demonstrate substantial improvements in MLLM accuracy on circuit understanding tasks, validating both the efficacy of our method and the utility of the new dataset.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Planning, Routing, and Scheduling: Planning with Language Models
• Computer Vision: Large Vision Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Large language models for search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Current multimodal large language models (MLLMs) struggle to understand circuit schematics due to their limited recognition capabilities. This could be attributed to the lack of high-quality schematic-netlist training data. Existing work such as AMSnet applies schematic parsing to generate netlists. However, these methods rely on hard-coded heuristics and are difficult to apply to complex or noisy schematics in this paper. We therefore propose a novel net detection mechanism based on segmentation with high robustness. The proposed method also recovers positional information, allowing digital reconstruction of schematics. We then expand AMSnet dataset with schematic images from various sources and create AMSnet 2.0. AMSnet 2.0 contains 2,686 circuits with schematic images, Spectre-formatted netlists, OpenAccess digital schematics, and positional information for circuit components and nets, whereas AMSnet only includes 792 circuits with SPICE netlists but no digital schematics.

Problem

• Improving MLLMs' understanding of circuit schematics
• Overcoming limitations of heuristic-based netlist generation
• Enhancing robustness and accuracy in net detection

Innovation

• AI segmentation for robust net detection
• Recovers positional information for digital reconstruction
• Expands dataset with diverse schematic sources