Defect Identification System Using Deep Neural Network for PCB Boards

Automatic Optical Inspection (AOI) systems are critical for boosting PCB production yield, but they frequently create high false alarm rates, necessitating costly manual inspections at Verification and Repair System (VRS) stations. To overcome this issue, we present an Automatic Defect Verification System (Auto-VRS) that uses Deep Neural Networks (DNNs) to accurately classify defects. Auto-VRS classifies PCB images as defective or non-defective and detects particular flaws such as short circuits, open circuits, and other irregularities. The system has a simple Graphical User Interface (GUI) that allows users to upload PCB images, preprocess data, classify problems, and view findings. Real-time flaw identification, defect region highlighting, and feedback on classification accuracy are among the key features. Key Words: PCB defect detection, deep neural networks, automatic defect verification, image classification, AOI system, machine vision, real-time inspection.