Research on PCB Defect Detection Method Based on Improved YOLOv8 Network

Printed Circuit Boards are crucial components in modern electronic products, widely used in computers, household appliances, automobiles, and medical devices. As technology advances, the demand for PCB reliability continues to rise, making the efficiency and accuracy of PCB defect detection increasingly significant. This paper proposes a PCB defect detection scheme based on an improved YOLOv8 network. By introducing deformable convolution DCNv2 into the C2f module of the original YOLOv8 network, sampling control at a broader feature level is enhanced. Additionally, using the occlusion-aware attention mechanism MultiSEAM improves the resolution of feature maps, aiding in the detection of small targets. Finally, training and testing were conducted on a PCB dataset published by Peking University. Experimental results show that the improved algorithm achieved a mean Average Precision of 95.6% and reduced computational complexity by 11.1%, significantly outperforming other common object detection algorithms in terms of detection accuracy.