Table 2.

Summary of banknote datasets used for experiments in the previous studies (Ref: Reference(s), N/A: Not Available).

Category		Ref.	Currency Type	Output Description	Dataset Availability
Banknote Recognition	Single Currency Recognition	[3]	INR, AUD, EUR, SAR, USD	2 denominations for each of INR, AUD, EUR, and SAR. USD was not reported.	N/A
		[4]	USD, RMB, EUR	24 classes of USD, 20 classes of RMB, and 28 classes of EUR.	N/A
		[5]	USD, Angola (AOA), Malawi (MWK), South Africa (ZAR)	68 classes of USD, 36 classes of AOA, 24 classes of MWK, and 40 classes of ZAR.	N/A
		[6]	Hongkong (HKD), Kazakhstan (KZT), Colombia (COP), USD	128 classes of HKD, 60 classes of KZT, 32 classes of COP, and 68 classes of USD.	N/A
	Multiple Currency Recognition	[7]	Japan (JPN), Italy (ITL), Spain (ESP), France (FRF)	23 denominations.	N/A
		[8]	KRW, USD, EUR, CNY, RUB	55 denominations.	N/A
		[9]	CNY, EUR, JPY, KRW, RUB, USD	248 classes of 62 denominations.	DMC-DB1 [9]
		[10]	23 countries (USD, RUB, KZT, JPY, INR, EUR, CNY, etc.)	101 denominations.	N/A
Banknote Fitness Classification		[11]	INR, KRW, USD	5 classes with 3 classes of case 1 (fit, normal and unfit) and 2 classes of case 2 (fit and unfit).	DF-DB2 [11]
		[12]	EUR, RUB	2 classes (fit and unfit).	N/A
		[14]	USD, KRW, INR	2 classes (fit and unfit).	N/A
		[15]	KRW, INR, USD	3 classes of KRW and INR (fit, normal, and unfit), 2 classes of USD (fit and unfit).	DF-DB1 [15]
Banknote type and fitness classification (proposed method)			INR, KRW, USD	116 classes of banknote kinds and fitness levels.	DF-DB3 [20]