PCB Defect Detection based on Generative Adversarial Network

Abstract

This paper proposes a PCB defect detection scheme based on the generative confrontation network, which can be applied to the automatic detection system of PCB vision inspection (vision inspection). We use the edge-enhanced super-resolution GAN (EESRGAN) applied in the field of remote sensing to enhance the PCB images and complete the super-resolution detection of the reconstructed picture. And use the PCB pictures of different preprocessing models in an end-to-end manner to compare the recognition of PCB defects after training. Experiments on the PCB data set show that the PCB pictures after sliding cutting are input into the result of EESRGAN training, which can relatively accurately identify the 6 types of defects contained in the data set. Our results show the effectiveness of our data processing methods.