SP-YOLO: A Model for Detecting Solder Paste Printing-Defect in PCB

As the cornerstone of the electronics manufacturing industry, printed circuit boards (PCBs) often suffer from defects caused by solder paste printing issues during the soldering process. However, traditional defect detection methods for solder paste printing face challenges such as poor recognition performance, low efficiency, and high costs. To address these issues, this paper proposes the SP-YOLO model, which can be deployed on Automated Optical Inspection (AOI) equipment for solder paste defect detection. The model introduces DAA into the backbone feature network and integrates it with convolution operations to enhance the ability to capture defect-related features. Experimental results show that SP-YOLO achieves a defect recognition accuracy of 95.1% in PCB solder paste printing inspection. Compared with the YOLOv11s model, its mAP is improved by 2.22%, demonstrating superior detection performance for difficult-to-identify defects. This provides significant application value for quality control in the electronics manufacturing industry.