Table 3 Experimental results on the crack dataset.

From: FCN attention enhancing asphalt pavement crack detection through attention mechanisms and fully convolutional networks

Model

Loss rate

(%)

Global accuracy rate

(%)

Average intersection and

union ratio (%)

Test duration

(ms)

VGG-16

39.49

78.864

53.1

139.12

AlexNet

27.86

84.971

57.4

143.82

FCN-attention

10.18

91.792

68.9

112.47

Mobile

37.65

83.351

51.5

120.98

GoogleNet

36.71

84.652

52.7

111.18

Back to article page