Table 6 Results of ablation experiments

From: Convolutional neural network based on transfer learning for discriminating the fermentation degree of black tea

Network

PolyLoss

AdamW

AT

FLOPs

Params

Accuracy

Precision

Recall

F1

FPS

ResNet18

   

1.824G

11.178M

0.9037

0.9065

0.8153

0.8519

75.24

√

  

0.9265

0.9164

0.8630

0.8836

73.75

√

√

 

0.9425

0.9272

0.8881

0.9064

74.60

√

√

√

0.9452

0.9280

0.9055

0.9164

74.22