PCB Surface Defect Detection Algorithm based on SC-YOLO

Printed circuit boards (PCBs) are fundamental components in electronic devices. This study introduces a YOLOv8 - based network for PCB defect detection. The defect features can be effectively extracted, which could lead to information loss reduced by the proposed improved C2f-SC module. Subsequently, we constructed the feature fusion structure based on the GSConv. By operating through independent spatial channel grouping convolution and cross-stage cascading, which can reduce network parameters and improves detection efficiency. CARAFE up-sampling method expands the receptive field by aggregating contextual information. In this research, the Sony zve - 10L was employed as the image acquisition device. The experimental outcomes demonstrate that the proposed model achieves a 4.2 % increase in mAP compared to YOLOv8. With only 2.77M parameters, the model offers a viable solution for real - time PCB defect detection in industrial applications.