A Novel Approach for Automatic Detection of Concrete Surface Voids Using Image Texture Analysis and History-Based Adaptive Differential Evolution Optimized Support Vector Machine

Table 2

Prediction performance of the L-SHADE-SVM-SVD.

Phases	Indices	CAR (%)	TP	TN	FP	FN	Precision	Recall	NPV	F1 score
Training	Mean	97.417	437.200	439.550	9.550	13.700	0.979	0.970	0.970	0.974
	Std.	0.222	4.938	5.753	1.605	1.302	0.004	0.003	0.003	0.002
Testing	Mean	92.600	46.200	46.400	4.500	2.900	0.911	0.942	0.940	0.926
	Std.	2.761	4.873	5.305	1.878	2.125	0.037	0.040	0.044	0.027