pid-parser: Turning P&ID Drawings Into a Graph With a Vision LLM
An open-source agent skill that reads a P&ID image and writes tables and a graph you can load into SQL or Neo4j. No GPU, no training: a vision LLM does the reading. With gpt-6.1-sol it scores a connection F1 of 0.80 on real PID2Graph sheets and 0.83 on synthetic ones, in three to six minutes a sheet.
aiagentscomputer-visionprocess-engineering
13 min read