Printed Circuit Board Defect Detection Based on Improved YOLOv5

In view of the problems such as insufficient accuracy and low efficiency in the detection of small defects on the surface of printed circuit boards, this study proposes a defect detection method based on the improved YOLOv5 algorithm. This method introduces the Swin-Transformer architecture into the backbone network, effectively integrating multi-scale feature information of local details and global context. It uses ECIoU_Loss as the loss function to achieve the coordinated optimization of detection speed and accuracy. At the same time, by integrating the Biformer attention mechanism and conducting iterative training on the printed circuit board dataset, an optimal detection model is finally obtained. The experimental results show that this system can achieve accurate identification and real-time positioning of defective circuit boards, with the [email protected] (Mean Average [email protected]) index reaching 0.991, significantly reducing the error rate of manual detection. Compared with existing mainstream detection models, the improved YOLOv5 model exhibits more excellent performance in the classification and positioning of circuit board defects.