Research on PCB Defect Detection Algorithm Based on Improved YOLOv8

In the electronics manufacturing industry, quality control of printed circuit boards (PCBs) is critical to guaranteeing the performance and reliability of the final product. Aiming at the problems that the defects of printed circuit boards (PCBs) are relatively small and difficult to be accurately identified, as well as the existing PCB defect detection algorithms have large computation and parameter counts, which can not meet the requirements of lightweight deployment, we propose a PCB defect detection method based on the improved YOLOv8. Firstly, Partial Convolution (PConv) is combined with the C2f module to propose C2f_PConv_Block, which reduces the computational volume of the model and the number of parameters while extracting the features more efficiently. Secondly, Bidirectional Feature Pyramid Network (BiFPN) is used in the feature fusion stage to achieve multi-scale feature extraction and fusion. In addition, RFB modules are added in front of each of the three decoupling heads to expand the receptive field and improve the accuracy of detection of small defects in the PCB. At the same time, the EIoU Loss is used instead of the CIoU Loss, which accelerates the speed of convergence and improves the accuracy of the regression. The experimental results show that the improved YOLOv8 detection algorithm has the mean average precision (mAP) of 95.5%, which is 4.2% higher than that of the original YOLOv8 algorithm, and at the same time the model computation and parameter counts are decreased by 12.3% and 10.0%, respectively. Compared with other advanced PCB defect detection algorithms, this paper's algorithm improves the detection accuracy while reduces the complexity of the model, and achieves lightweight, high-precision PCB defect detection.