ChiShengChen / Food2K-TW101

Food2K-TW101

All the accuracy (ACC) are the trained models do inference on un-seen validation data, not the training accuracy.

Dataset	Model	Epochs	Optimizer	Top-1 ACC	Top-5 ACC	Pretrain	Augmentation type
FOOD2K	PRENet-ResNet50	N/A	ongoing	83.03	97.21	ongoing	ongoing
FOOD2K	Inception V4	N/A	ongoing	82.02	96.45	ongoing	ongoing
FOOD2K	VGG16	N/A	ongoing	78.96	95.26	ongoing	ongoing

Dataset	Model	Epochs	Optimizer	Top-1 ACC	Top-5 ACC	Pretrain	Augmentation type
Taiwanese Food-101	MobileFormer_508M	100	AdamW	95.15	99.52	imagenet-1k	Basic
Taiwanese Food-101	PRENet-ResNet50	100	SGD	94.59	99.49	food2K	Basic
Taiwanese Food-101	PVTv2-B2-Linear	100	AdamW	94.51	99.45	imagenet-1k	Basic
Taiwanese Food-101	ConvNeXtv2_nano	100	AdamW	94.04	99.29	imagenet-1k	Basic
Taiwanese Food-101	ResNet50	100	AdamW	93.96	99.58	imagenet-1k	Basic
Taiwanese Food-101	ConvNeXtv2_femto	100	AdamW	93.21	99.19	imagenet-1k	Basic
Taiwanese Food-101	MobileFormer_96M	100	AdamW	93.01	99.25	imagenet-1k	Basic
Taiwanese Food-101	Inception V4	100	SGD	92.14	99.01	imagenet-1k	Basic
Taiwanese Food-101	mobilenetv3_large_100	100	AdamW	91.66	98.97	imagenet-1k	Basic
Taiwanese Food-101	mobileViT v2	100	AdamW	89.98	98.46	imagenet-1k	Basic
Taiwanese Food-101	efficientNetv3 Large	100	AdamW	84.89	96.40	imagenet-1k	Basic
Taiwanese Food-101	efficient ViT_MIT	100	AdamW	82.32	95.78	imagenet-1k	Basic
Taiwanese Food-101	RepViT_m2.3	100	AdamW	76.53	93.80	imagenet-1k	Basic
Taiwanese Food-101	RepViT_m0.9	100	AdamW	75.01	93.49	imagenet-1k	Basic
Taiwanese Food-101	VGG16	N/A	SGD	67.65	89.33	imagenet-1k	Basic

Dataset	Model	Epochs	Optimizer	Top-1 ACC	Top-5 ACC	Pretrain	Augmentation type
FOOD2K-TW	PRENet-ResNet50	N/A	ongoing	ongoing	ongoing	ongoing	ongoing
FOOD2K-TW	Inception V4	100	SGD	81.43	96.28	imagenet-1k	Basic
FOOD2K-TW	VGG16	N/A	ongoing	ongoing	ongoing	ongoing	ongoing

Do not trainable

Dataset	Model	Epochs	Optimizer	Top-1 ACC	Top-5 ACC	Pretrain	Augmentation type
Taiwanese Food-101	convnextv2_huge.fcmae_ft_in1k	15	AdamW	1.02	0.99	imagenet-1k	Normal

Augmentation type

Type	Detail
Basic	RandomHorizontalFlip(p=0.5) + RandomRotation(degrees=15) + ColorJitter(brightness=0.126, saturation=0.5) + Resize((550, 550)) + RandomCrop(448)

To-Do List

Dataset	Classes/Images	paper	dataset	Data Aviliable?
Taiwanese-Food-101	101/20,200	N/A (only master Thesis)	dataset	Yes
UEC Food256	256/25,088	paper	dataset	Yes
ETH Food-101	101/101,000	paper	dataset	Yes
Vireo Food-172	172/110,241	paper	dataset	Need to email
Food524DB	524/247,636	paper	dataset	Yes, but .mat format
CNFOOD-241	241/191,811	paper	dataset	Yes
ChineseFoodNet	208/192,000	paper	dataset	Link is dead, email didnot response
Sushi-50	50/3,963	paper	dataset	Yes
ISIA Food-500	500/399,726	paper	dataset	Yes
Food2K	2,000/1,036,564	paper	dataset	Need to email. Yes

TaiwaneseFood101 https://www.kaggle.com/datasets/kuantinglai/taiwanese-food-101/data

To-Play List

https://arxiv.org/abs/2301.10936 https://www.kaggle.com/datasets/zachaluza/cnfood-241 https://universe.roboflow.com/search?q=class%3Adumpling https://universe.roboflow.com/search?q=class%253Apork&p=1 https://ieeexplore.ieee.org/document/9214438 https://www.researchgate.net/publication/372133738_Deep_Learning_for_Food_Image_Recognition_and_Nutrition_Analysis_Towards_Chronic_Diseases_Monitoring_A_Systematic_Review https://ndltd.ncl.edu.tw/cgi-bin/gs32/gsweb.cgi/login?o=dnclcdr&s=id=%22111NYPI0294006%22.&searchmode=basic