Table 6 The 5-fold comparison results.

From: Optimizing deep belief network for concrete crack detection via a modified design of ideal gas molecular dynamics

Method

Accuracy (%)

Specificity (%)

Fl Score

Precision (%)

Recall (%)

FCN9

84.442

82.571

75.139

80.767

78.8

YOLO10

83.911

87.414

79.687

84.862

86.325

CrackSegNet11

78.039

79.527

73.525

80.476

88.188

CNN12

80.325

80.486

78.906

88.792

93.311

CedNet13

80.259

84.323

77.797

89.542

95.11

DBN/MIGMM

90.189

94.502

88.093

94.586

94.529

Back to article page