Table 3.

Performance comparison on STARE.

Methods		Acc (%)		Spe (%)		Sen (%)		AUC (%)
		Base	Impro	Base	Impro	Base	Impro	Base	Impro
Proposed	5-way 3-shot	94.59	95.63	96.71	97.39	77.70	81.53	95.63	96.92
	4-way 3-shot	94.46	95.92	96.50	98.16	78.18	78.04	95.91	97.29
	3-way 3-shot	94.89	95.81	97.06	97.73	77.56	80.47	96.06	97.16
	3-way 4-shot	94.88	95.77	97.33	97.69	75.39	80.46	95.97	97.25
	3-way 5-shot	94.63	95.65	96.64	97.19	78.51	83.37	96.19	97.53
Machine learning	Fraz et al. (14)	95.34		97.63		75.48		97.68
	Aslani and Sarnel (16)	96.05		98.37		75.56		97.89
	Orlando et al. (17)	-		97.38		76.80		-
	Li et al. (18)	96.28		98.44		77.26		98.79
	Srinidhi et al. (19)	95.02		97.46		83.25		96.70
	Liskowski and Krawiec (22)	97.29		98.62		85.54		99.28
	Mo and Zhang (23)	96.74		98.44		81.47		98.85
	Jiang et al. (24)	97.34		98.46		83.52		99.00
	Zhou et al. (25)	95.85		97.61		80.65		-
	Yan et al. (26)	96.12		98.46		75.81		98.01
	Filipe et al. (27)	96.94		98.58		83.15		99.05
	Park et al. (28)	98.76		99.38		83.24		98.73

The bold values indicate the maximum value of the corresponding metric in the manuscript.