Neural Network Classification of Printed Circuit Board Defects

This study focuses on the development of a model based on the YOLOv8x neural network for the automated detection of defects in printed circuit boards (PCBs). Purpose: to train a neural network capable of effectively detecting and classifying various types of PCB defects. Methods: a deep learning method based on the YOLOv8x architecture, designed for object detection tasks. An accuracy and loss metrics analysis was conducted to assess the model’s effectiveness. Results: the trained model demonstrates high accuracy in classifying defects such as missing holes, shorts, and spurious copper, achieving an accuracy of 1.0. The “open circuits” and “copper spurs” classes also yield satisfactory results. However, the “mouse bites” class requires further improvement. Practical significance: the potential application of the developed model for automating quality control processes in PCBs could significantly enhance the reliability of electronic devices and reduce the potential failures in critical systems.