Printed Circuit Board Defect Detection Using Generative Deep Learning Model

Abstract

Recently, Printed Circuit Board(PCB) design of complex structures is essential to design small electronic devices. Defect detection is one of the most important PCB processes because PCB defects have a fatal effect on product quality. For previous defect detection, methods such as Automated Optical Inspection(AOI) or In-Circuit Test(ICT) were used. But these methods have their disadvantage. They need high-cost inspection equipment, and new setting values are required each time as the surrounding environment changes. The proposed system is robust to environmental changes using generative deep learning models. In addition, it is more convenient to use than existing deep learning models using semi-supervised learning.