2020/07/20
UECFoodPix(GrabCut ver) dataset file is released.
UECFoodPixComplete(Manual ver) dataset file will be released at a later date.
2020/10/22
UECFoodPixComplete(Manual ver) dataset is released.
UECFoodPix and UECFoodPixComplete are food images dataset with segmentation masks including 9,000 images for training and 1,000 image for testing. The Segmentation masks are augmented by food category. In UECFoodPix, the mask images is created using a bounding box and GrabCut, and In UECFoodPixComplete there is provided manualy.The mask images have pixel-wise food 103 class labels, and only R(red) channel have this labels.See Labels below for the correspondence between mask labels and pixel values.
UECFoodPixComplete Dataset (759 MB)
[LINK] UEC-Food100/256 dataset
This tar file contains the following files.
UECFOODPIXCOMPLTE/
├data/
└UECFoodPIXCOMPLTE/
└train/
│ └img/ .jpg
│ └mask/ .png
└test/
│ └img/ .jpg
│ └mask/ .png
└category.txt
└train.txt
└test.txt
In “img”, real images are saved as “jpg”.
1.jpg
10.jpg
100.jpg
1000.jpg
In “mask”, mask images are saved as “png”.
1.png
10.png
100.png
1000.png
In “category.txt”, class labels are contained.
id name
1 rice
2 eels on rice
3 pilaf
In “train.txt” and “test,txt”, the value of the image name without extension contained.
1227
1257
1287
1305
Name | RGB | Name | RGB | ||
---|---|---|---|---|---|
0 | ‘background’ | 0,0,0 | 20 | ‘udon noodle’ | 20,0,0 |
1 | ‘rice’ | 1,0,0 | 21 | ‘tempura udon’ | 21,0,0 |
2 | ‘eels on rice’ | 2,0,0 | 22 | ‘soba noodle’ | 22,0,0 |
3 | ‘pilaf’ | 3,0,0 | 23 | ‘ramen noodle’ | 23,0,0 |
4 | ‘chicken-’n’-egg on rice’ | 4,0,0 | 24 | ‘beef noodle’ | 24,0,0 |
5 | ‘pork cutlet on rice’ | 5,0,0 | 25 | ‘tensin noodle’ | 25,0,0 |
6 | ‘beef curry’ | 6,0,0 | 26 | ‘fried noodle’ | 26,0,0 |
7 | ‘sushi’ | 7,0,0 | 27 | ‘spaghetti’ | 27,0,0 |
8 | ‘chicken rice’ | 8,0,0 | 28 | ‘Japanese-style-pancake’ | 28,0,0 |
9 | ‘fried rice’ | 9,0,0 | 29 | ‘takoyaki’ | 29,0,0 |
10 | ‘tempura bowl’ | 10,0,0 | 30 | ‘gratin’ | 30,0,0 |
11 | ‘bibimbap’ | 11,0,0 | 31 | ‘sauteed vegetables’ | 31,0,0 |
12 | ‘toast’ | 12,0,0 | 32 | ‘croquette’ | 32,0,0 |
13 | ‘croissant’ | 13,0,0 | 33 | ‘grilled eggplant’ | 33,0,0 |
14 | ‘roll bread’ | 14,0,0 | 34 | ‘sauteed spinach’ | 34,0,0 |
15 | ‘raisin bread’ | 15,0,0 | 35 | ‘vegetable tempura’ | 35,0,0 |
16 | ‘chip butty’ | 16,0,0 | 36 | ‘miso soup’ | 36,0,0 |
17 | ‘hamburger’ | 17,0,0 | 37 | ‘potage’ | 37,0,0 |
18 | ‘pizza’ | 18,0,0 | 38 | ‘sausage’ | 38,0,0 |
19 | ‘sandwiches’ | 19,0,0 | 39 | ‘oden’ | 39,0,0 |
Name | RGB | Name | RGB | ||
---|---|---|---|---|---|
40 | ‘omelete’ | 40,0,0 | 60 | ‘hambarg steak’ | 60,0,0 |
41 | ‘ganmodokii’ | 41,0,0 | 61 | ‘beef steak’ | 61,0,0 |
42 | ‘jiaozi’ | 42,0,0 | 62 | ‘dried fish’ | 62,0,0 |
43 | ‘stew’ | 43,0,0 | 63 | ‘ginger pork saute’ | 63,0,0 |
44 | ‘teriyaki grilled fish’ | 44,0,0 | 64 | ‘spicy chili-flavored tofu’ | 64,0,0 |
45 | ‘fried fish’ | 45,0,0 | 65 | ‘yakitri’ | 65,0,0 |
46 | ‘grilled salmon’ | 46,0,0 | 66 | ‘cabbage roll’ | 66,0,0 |
47 | ‘salmon meuniere’ | 47,0,0 | 67 | ‘rolled omelete’ | 67,0,0 |
48 | ‘sashimi’ | 48,0,0 | 68 | ‘ege sunny-side up’ | 68,0,0 |
49 | ‘grilled pacific saury’ | 49,0,0 | 69 | ‘fermented soybeans’ | 69,0,0 |
50 | ‘sukiyaki’ | 50,0,0 | 70 | ‘cold tofu’ | 70,0,0 |
51 | ‘sweet and sour pork’ | 51,0,0 | 71 | ‘egg roll’ | 71,0,0 |
52 | ‘lightly roasted fish’ | 52,0,0 | 72 | ‘chilled noodle’ | 72,0,0 |
53 | ‘steamed egg hotchpotch’ | 53,0,0 | 73 | ‘stir-fried beef and pappers’ | 73,0,0 |
54 | ‘tempura’ | 54,0,0 | 74 | ‘simmered pork’ | 74,0,0 |
55 | ‘fried chicken’ | 55,0,0 | 75 | ‘boiled chicken and vegetables’ | 75,0,0 |
56 | ‘sirloin cutlet’ | 56,0,0 | 76 | ‘sashimi bowl’ | 76,0,0 |
57 | ‘nanbanzuke’ | 57,0,0 | 77 | ‘sushi bowl’ | 77,0,0 |
58 | ‘boiled fish’ | 58,0,0 | 78 | ‘fish-shaped panceke with bean jam’ | 78,0,0 |
59 | ‘seasoned beef with potatoes’ | 59,0,0 | 79 | ‘shrimp with chill source’ | 79,0,0 |
Name | RGB | |
---|---|---|
80 | ‘roast chicken’ | 80,0,0 |
81 | ‘steamed meat dumpling’ | 81,0,0 |
82 | ‘omelet with fried rice’ | 82,0,0 |
83 | ‘cutlet curry’ | 83,0,0 |
84 | ‘spaghetti meat sauce’ | 84,0,0 |
85 | ‘fried shrimp’ | 85,0,0 |
86 | ‘potato salad’ | 86,0,0 |
87 | ‘green salad’ | 87,0,0 |
88 | ‘macaroni salad’ | 88,0,0 |
89 | ‘Japanese tofu and vegetable chowder’ | 89,0,0 |
90 | ‘pork miso soup’ | 90,0,0 |
91 | ‘chinese soup’ | 91,0,0 |
92 | ‘beef bowl’ | 92,0,0 |
93 | ‘kinpira-style sauteed burdock’ | 93,0,0 |
94 | ‘rice ball’ | 94,0,0 |
95 | ‘pizza toast’ | 95,0,0 |
96 | ‘dipping noodels’ | 96,0,0 |
97 | ‘hot dog’ | 97,0,0 |
98 | ‘french fries’ | 98,0,0 |
99 | ‘mixed rice’ | 99,0,0 |
100 | ‘goya chanpuru’ | 100,0,0 |
101 | ‘others’ | 101,0,0 |
102 | ‘beverage’ | 102,0,0 |
The results of segmentation by using Deep lab v3+
Accuracy | mIoU | |
---|---|---|
UECFOODPIX | 0.560 | 0.416 |
UECFOODPIXCOMPLETE | 0.668 | 0.555 |
UECFoodPix and UECFoodPixComplete dataset can be used only for non-commercial research purpose.
If you publish a paper using our dataset, we’d glad if you could refer to the following paper:
UECFoodPix
@inproceedings{uecfoodpix,
title={A New Large-scale Food Image Segmentation Dataset and Its Application to Food Calorie Estimation Based on Grains of Rice},
author={Ege, Takumi and Yanai, Keiji},
booktitle={Proc. of ACMMM Workshop on Multimedia Assisted Dietary Management(MADiMa)},
year={2019},
}
UECFoodPixComplete [PDF]
@inproceedings{uecfoodpixcomplete,
title={{UEC-FoodPIX Complete}: A Large-scale Food Image Segmentation Dataset},
author={Okamoto, Kaimu and Yanai, Keiji},
booktitle={Proc. of ICPR Workshop on Multimedia Assisted Dietary Management(MADiMa)},
year={2021},
}
Please feel free to contact us with any comments and questions by email
(Kaimu Okamoto, Prof. Keiji Yanai)