[Retracted] X-Ray Small Target Security Inspection Based on TB-YOLOv5

Table 1

Four network structures of YOLOv5. The parameters depth multiple and width multiple control the number of BottleneckCSP and the number of convolution kernels of the architectural models, respectively.

	YOLOv5-s	YOLOv5-m	YOLOv5-l	YOLOv5-x
Depth multiple	0.33	0.67	1.0	1.33
Width multiple	0.50	0.75	1.0	1.25
BottleneckCSP BCSPn (true)	1, 3, 3	2, 6, 6	3, 9, 9	4, 12, 12
BottleneckCSP BCSPn (false)	1	2	3	4
Number of convolution kernels	32, 64, 128, 256, 512	48, 96, 192, 384, 768	64, 128, 256, 512, 1024	80, 160, 320, 640, 1280