Table 17.

Class-wise performance analysis of ResNet50.

Class	NAC a	NCC b	Accuracy	SD c	Precision	Recall	F1-score
Original	3000	3046	99.81%	0.04	0.98	1.0	0.99
Gaussian	3000	4021	87.26%	3.19	0.49	0.66	0.56
Localvar	3000	1974	87.28%	2.45	0.49	0.32	0.39
Pepper	3000	3002	99.99%	0.00	1.0	1.0	1.0
Poisson	3000	2941	99.75%	0.09	1.0	0.98	0.99
Salt&Pepper	3000	3006	99.97%	0.01	1.0	1.0	1.0
Salt	3000	3053	99.78%	0.07	0.98	1.0	0.99
Speckle	3000	2957	99.82%	0.03	1.0	0.99	0.99

^aNumber of actual cases in a class.

^bNumber of cases classified as belonging to a class.

^cStandard Deviation.