Image
Recyclable waste detection
BBOX
Environment
Due to COVID-19, delivery services have experienced exponential growth which generated vast amount of waste, to the point where preexisting waste disposal facilities ran out of capacity. This project started from acknowledging such difficulties. The aim is to develop AI technology that could recognize and classify wastes, which is the basis for automated waste disposal and classification.
Datumo and Dingbro have cooperated in building this dataset to further the research on domestic waste recognition, as part of the 2021 AI Grand Challenge held by the Korean Ministry of Science and ICT.
Dingbro has put much effort in solving problems of the society via AI, which led them to build bounding box datasets of domestic waste to develop technology that could reduce waste problems. The dataset has allowed to majorly improve the accuracy of the company’s object detection model, which won 1st place in the 2021 AI Grand Challenge-Domestic Waste Recognition Challenge, surpassing prestigious research labs and enterprises. By deciding to openly publish this dataset, Dingbro hopes it could help waste recognition problems.
We hope that the publication of this data set will help the area of waste recognition issues.
“Taking too much time and having uneven quality were the two toughest issues in labeling data.
Datumo, however, advised us on writing manuals and helped us design roadmaps for making decisions in various possible situations. Datumo's experiences in previous projects helped us to set guidelines for a variety of possibilities which resulted in satisfying quality of data labeling.
We requested quite amount of data and yet received it much earlier than we expected, thanks to Datumo's crowd-sourcing system. The final inspection process was very strict, which was extremely appreciated by our researchers. The finalized data were in desired format to train our AI directly, without additional processing. We would like to thank Datumo with all our hearts for providing much help to bring our project to reality. ”
*: Spy system is Datumo's worker monitoring system. Workers with low labeling accuracy are classified as 'spies' and restrictions are placed on specific tasks or projects.
Of the total data collected, the label discloses 100% of the data that exists.
Image including 1~5 domestic waste (36,735 images)
JSON file including waste image path and label information (36,735 files)
label Enum list
- 'paper' (종이) - 'pack' (종이팩) - 'can' (캔,철류) - 'glass' (유리) - 'pet' (페트) - 'plastic' (플라스틱) - 'vinyl' (비닐)
Data collection and labeling were completed using the mobile version of Datumo's crowd-sourcing platform, Cash Mission.
Data were collected and labeled using Cash Mission, Datumo's crowd-sourcing platform.
Sample Data
{
"version": "5.0.1",
"flags": {},
"shapes": [
{
"label": "pet",
"points": [
[
969.5652173913043,
1498.2608695652173
],
[
1164.3478260869563,
2026.0869565217388
]
],
"group_id": null,
"shape_type": "rectangle",
"flags": {}
}
],
"imagePath": "3667931.jpg",
"imageData": null,
"imageHeight": 4032,
"imageWidth": 2268
}
{
"version": "5.0.1",
"flags": {},
"shapes": [
{
"label": "can",
"points": [
[
937.5000000000003,
1856.25
],
[
1221.8750000000005,
2143.75
]
],
"group_id": null,
"shape_type": "rectangle",
"flags": {}
},
{
"label": "vinyl",
"points": [
[
1228.1250000000002,
1346.8750000000002
],
[
1665.6250000000005,
2162.5
]
],
"group_id": null,
"shape_type": "rectangle",
"flags": {}
},
{
"label": "can",
"points": [
[
568.75,
1287.5
],
[
1296.875,
2078.125
]
],
"group_id": null,
"shape_type": "rectangle",
"flags": {}
}
],
"imagePath": "4072172.jpg",
"imageData": null,
"imageHeight": 4000,
"imageWidth": 2250
}
{
"version": "5.0.1",
"flags": {},
"shapes": [
{
"label": "plastic",
"points": [
[
1539.2280471206914,
871.7954817708764
],
[
1889.1611417431513,
1365.0540338019357
]
],
"group_id": null,
"shape_type": "rectangle",
"flags": {}
},
{
"label": "vinyl",
"points": [
[
1949.5235394363988,
800.6726590480359
],
[
2324.1314192021073,
1571.9488443491368
]
],
"group_id": null,
"shape_type": "rectangle",
"flags": {}
}
],
"imagePath": "4162461.jpg",
"imageData": null,
"imageHeight": 2268,
"imageWidth": 4032
}Environmental business or research regarding domestic waste classification
CC BY-SA
Reusers are allowed to distribute, remix, adapt, and build upon the material in any medium or format, even commercially, so long as attribution is given to the creator. If you remix, adapt, or build upon the material, you must license the modified material under identical terms.
https://creativecommons.org/licenses/by-sa/3.0/deed.en