Table 2.

Comparison of our algorithm with several popular models.

Method	Parameters (M)	Accuracy (%)	Efficiency (MFLOPs)
AlexNet [21]	57.1	98.2	1198.7
GoogLeNet [14]	5.6	99.4	2589.1
MobileNet V2 [15]	2.2	98.7	587.5
ResNet-34 [13]	21.2	99.2	6849.5
DenseNet-121 [22]	6.9	98.2	4991.4
ShuffleNet V2 [15]	0.35	97.3	78.6
Ours	0.17	99.4	211.9