Table 2 Shows the comparison of Param, FLOPs on COVID-19 DataSet.

From: Medical image segmentation model based on triple gate MultiLayer perceptron

Model

Backbone

Param (M)

FLOPs (G)

PSPNet34

ResNet50

64.03

257.79

DeepLabv315

ResNet50

38.71

163.83

DFN41

ResNet50

43.53

81.88

EncNet42

ResNet50

51.25

217.46

OCNet43

ResNet50

51.60

220.69

DANet44

ResNet50

64.87

275.72

Inf-Net35

ResNet50

30.19

27.30

UNet++12

-

8.95

138.37

Attention U-Net13

-

8.52

67.14

TGMLP U-Net(ours)

-

3.03

1.9

  1. The best values of each indicator are in bold. (“-” indicates that the model does not use the classical deep learning network as the backbone network.)
Back to article page