The goal of this project is to apply deep learning methods and image processing techniques to automatically identify defects such as incorrect soldering, missing components, or faulty connections, improving the quality inspection process for PCBs

Intelligent PCB anomaly detection system based on OpenCV. Features quadrant-aware direction adaptive cutting, pad masking constraints, and structural affinity scoring for precise coordinate generation. [Under Continuous Optimization]

Agentic PCB defect inspection using Amazon Nova on AWS Bedrock — 7-step MCP pipeline with autonomous tool-calling (quarantine, work orders, knowledge graph)

Production PCB defect detection: 99.5% mAP, 3x inference speedup (ONNX+INT8), FastAPI deployment. YOLOv8 + complete MLOps pipeline.