TABLE 1.

Publicly available food datasets used as input to image-based food recognition systems

No.	Name	Year	Food categories, n	Total number of images, n	Food items in each image	Cuisine	Reference
1	Pittsburgh Fast-food Image Dataset (PFID)	2009	61	1089	Single	Fast food	(23)
2	UEC-Food 100	2012	100	10,000	Single and multiple	Japanese	(24)
3	NTU-FOOD	2012	50	5000	Single	Multiethnic	(28)
4	UNICT-FD889	2014	889	3583	Single	Multiethnic	(26)
5	Food-101	2014	101	101,000	Single	Multiethnic	(29)
6	UEC-Food 256	2014	256	31,397	Single and multiple	Multiethnic	(25)
7	Ambient Kitchen	2014	12	1800	Single and multiple	Multiethnic	(30)
8	UPMC Food-101	2015	101	90,840	Single	Multiethnic	(31)
9	Dishes	2015	3832	117,504	Single	Multiethnic	(32)
10	Menu-Match	2015	41	646	Single and multiple	Asian, Italian	(33)
11	FooDD	2015	23	3000	Single and multiple	Multiethnic	(34)
12	UNIMIB 2015	2015	15	2000	Multiple	Italian	(35)
13	Instagram800K	2016	43	808,964	Single and multiple	Multiethnic	(36)
14	UNICT-FD1200	2016	1200	4754	Single and multiple	Multiethnic	(27)
15	UNIMIB 2016	2016	73	1027	Multiple	Italian	(6)
16	EgocentricFood	2016	9	5038	Multiple	Multiethnic	(17)
17	VIREO Food-172	2016	172	110,241	Single and multiple	Chinese	(37)
18	FOOD-5K	2016	2	5000	Multiple	Multiethnic	(38)
19	FOOD-11	2016	11	16,643	Multiple	Multiethnic	(38)
20	NTUA-Food 2017	2017	82	3248	Single	Multiethnic	(39)
21	ECUST Food Dataset	2017	19	2978	Single and multiple	Multiethnic	(40)
22	Madima 2017	2017	21	21,807	Multiple	Central European	(43)
23	Food524DB	2017	524	247,636	Single and multiple	Multiethnic	(41)
24	ChineseFoodNet	2017	208	192,000	Single and multiple	Chinese	(44)
25	Eating Occasion Image to Food Energy	2020	21	96	Multiple	Multiethnic	(45)
26	ChinaFood-100	2021	100	10,074	Single	Chinese	(22)
27	VIPER-FoodNet	2021	82	14,991	Multiple	Multiethnic	(46)