Skip to main content

Table 1 Camelyon16 segmentation results comparison

From: Large-scale tissue histopathology image segmentation based on feature pyramid

 

FCN-VGG16

U-Net

SegNet

Resnet50-GICN (k = 7)

Resnet50-GPP

Resnet50-GICN-GPP (k = 7) (proposed method)

Score

46.55

37.4

50.82

59.18

61.33

63.70

Params

134 M

7 M

31 M

832 K

2619 K

3176 K

Test time

210 ms

73 ms

143 ms

69 ms

77 ms

102 ms

Stride

32

16

32

8

16

8

  1. Score means segmentation accuracy (%). Params is the number of network model parameters. Test time means how long a 512 × 512 size image takes in the test. Stride means the maximum multiple of subsampling