Table 4 Comparison of computational complexity.

From: Hybrid gabor attention convolution and transformer interaction network with hierarchical monitoring mechanism for liver and tumor segmentation

Method

Params (M)

Flops (G)

Inference time (s)

DeepLabv3+

58.03

0.01747

5.62

U-Net

8.64

0.00260

2.40

Attention U-Net

8.73

0.00263

2.57

ResU-Net

31.56

0.00953

2.38

U-Net++

36.63

0.01103

2.99

Double UNet

29.29

0.00882

3.75

nnformer

58.32

0.02854

4.12

Swin-UNet

27.17

0.00818

3.36

TransUNet

105.28

0.03169

4.45

HiFormer

25.51

0.00768

3.51

HyborNet

28.55

0.01522

4.31

Back to article page