Research on a Lightweight PCB Detection Algorithm Based on AE-YOLO

Abstract

The attention enhancement YOLO printed circuit board (PCB) defect detection algorithm AE-YOLO, which improves YOLOv8, is proposed to improve the current slow detection speed of PCB defect detection problems, such as high missed detection or false detection rates and low detection accuracy. First, in the backbone network, CoT Net is used instead of the original feature extraction network to reduce the number of parameters of the model and improve its detection speed while maintaining the original detection accuracy as much as possible. Then, the SPPFS module is used in the last layer of the backbone network to enhance the model’s ability to extract global information, fuse global features, and use rich primary semantic information to pave the way for subsequent classification and positioning. Finally, the CC3 module is used to perceive high-level semantic information to help the decoupled detection head better perform target classification and prediction positioning, improve the detection accuracy and comprehensiveness of the model, and provide the model with continuous performance improvements. Compared with the original YOLOv8 model, the AE-YOLO algorithm compresses the parameters by 16%, increases the detection accuracy by 2.9%, and increases the recall rate by 3.3%. This algorithm provides a more efficient method for PCB defect detection.

Keywords

• CoT block
• lightweight network
• PCB defect detections
• YOLOv8