A dataset for the task of beer bottle localization by Christopher Kümmel, Robin Mehlitz and Tom Oberhauser from Beuth University of Applied Sciences Berlin.
We present BeerBB-1K, a dataset for the task of beer bottle localization. This dataset was generated by Christopher Kümmel, Robin Mehlitz and Tom Oberhauser in 2017 for a project in Interactive Systems by Prof. Dr.-Ing. Kristian Hildebrand at the Beuth University of Applied Sciences Berlin.
The dataset contains 1000 .jpeg images of a fixed size of 1000x750px. Every image has one or more beer bottles on it. Every image has a metadata .json file that contains information about the bounding boxes of the beer bottles. In addition to this it also contains information about the beer brand and the open-state of the bottle.
The pictures vary in quality and complexity to reflect the quality of real world smartphone images.
0000.jpg-0999.jpg- color images 1000x750px
0000.json-0999.json- contains an array of bounding box objects which contain:
x- x coordinate of upper left corner of the bounding boxy- y coordinate of upper left corner of the bounding boxw- width of the bounding boxh- height of the bounding boxisOpen- boolean, is the bottle open?brand- beer brand
- contains an array of bounding box objects which contain:
[{
"brand": "Berliner Kindl Pilsener",
"h": 445,
"isOpen": true,
"w": 107,
"x": 369,
"y": 100
}, {
"brand": "Radeberger Pilsner",
"h": 625,
"isOpen": false,
"w": 151,
"x": 422,
"y": 58
}, {
"brand": "Berliner Kindl Pilsener",
"h": 629,
"isOpen": false,
"w": 169,
"x": 582,
"y": 75
}]This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.