基于改进YOLOv5s的印刷电路板缺陷检测算法
周著国,鲁玉军,吕利叶

Improved YOLOv5s-based algorithm for printed circuit board defect detection
Zhuguo ZHOU,Yujun LU,Liye LV
表 3 不同特征融合网络植入后的模型检测性能对比
Tab.3 Comparison of model detection performance after implanting different feature fusion networks
特征融合方式P/%mAP/%FLOPs/109
BiFPN96.594.915.5
AFPN96.194.116.2
ASFF96.694.716.2
HSFPN97.195.214.5
GFPN97.595.616.1
GDFPN98.397.615.6