Table 5 Performance comparison results of different models.

From: A lightweight semantic segmentation method for concrete bridge surface diseases based on improved DeeplabV3+

Method

mIoU/%

mPA/%

Parameters/106

FPS

U-Net

69.86

76.62

38.74

40.41

HRNet

71.32

79.53

18.64

18.26

PSPNet

64.99

72.15

2.62

70.56

SETR

71.65

80.60

98.49

12.87

SegFormer

73.38

82.42

48.64

37.23

Mask2Former

60.94

69.34

44.07

24.61

PIDNet

73.20

82.18

7.67

45.03

DeeplabV3+

71.51

80.47

72.05

16.42

Ours

75.24

84.68

6.97

52.64

  1. Significant values are in [bold].
Back to article page