An Efficient Surface Defect Detection Method for Printed Circuit Board

Printed circuit board (PCB) surface defect detection is crucial for ensuring the reliability and stability of electronic products. Existing PCB defect detection methods suffer from high model complexity, slow detection speed, and are prone to missed and false detection for tiny defect targets. To address these challenges, this study proposes an improved PCB surface defect detection model based on YOLOv8n. Firstly, the method introduces an Edge Feature Enhancement (EFE) module, which consists of center difference convolution to better extract the edge information of defects. Additionally, VOV-GSCSP is introduced into the neck network to reduce model parameters while maintaining accuracy. Finally, we optimize the model's loss function with SIoU loss to further improve the convergence speed and detection accuracy. The experimental results demonstrate that the mean average precision (mAP) of the proposed algorithm reaches 96.5%, 1.7% higher than the original YOLOv8n model. In addition, the proposed model has only 2.9M parameters and a detection speed of 155 frames per second (FPS). It guarantees both detection efficiency and accuracy, and provides a feasible solution for deployment on mobile devices.