Automatic Impervious Surface Area Detection Using Image Texture Analysis and Neural Computing Models with Advanced Optimizers

Table 1

Training performance.

Models	Statistics	Performance measurement indices
		CAR (%)	TP	TN	FP	FN	Precision	Recall	NPV	F1 score
GDM	Mean	94.210	999.800	978.600	50.650	70.950	0.952	0.939	0.933	0.944
	Std	3.613	28.037	82.229	31.922	86.497	0.030	0.060	0.081	0.030
Adam	Mean	97.771	1,008.450	1,044.750	36.850	9.950	0.965	0.990	0.991	0.977
	Std	0.348	11.859	12.825	7.164	3.186	0.007	0.003	0.003	0.004
Adamax	Mean	97.340	1,008.500	1,035.650	42.700	13.150	0.959	0.987	0.987	0.973
	Std	0.392	18.629	15.415	5.780	3.991	0.006	0.004	0.004	0.004
Nadam	Mean	97.967	1,019.050	1,038.250	36.250	6.450	0.966	0.994	0.994	0.979
	Std	0.367	14.116	13.634	6.898	2.636	0.006	0.003	0.003	0.004
AdamW	Mean	97.793	1,014.100	1,039.550	40.050	6.300	0.962	0.994	0.994	0.978
	Std	0.310	12.992	12.698	6.289	3.480	0.006	0.003	0.003	0.003
AMSGrad	Mean	98.117	1,016.450	1,044.000	29.250	10.300	0.972	0.990	0.990	0.981
	Std	0.465	13.959	14.601	7.175	5.178	0.007	0.005	0.005	0.005