Mask Detection Dataset for yolo annotated by using https://github.com/tzutalin/labelImg
This dataset is in format for training yolov3 model using darknet or any python wrapper for yolo.
it has 4000+ images of masked people on and 4000 images of non-masks.
More data will be added in future and any contribution is appriciated.
Star this repo to stay updated.
class number | class name |
---|---|
0 | mask |
1 | nomask |
Directory structure:
img/
- image1.jpg
- image1.txt
- image2.jpg
- image2.txt
- image3.jpg
- image3.txt
.
.
.
train.txt
test.txt
format of annotation file
<object-class> <x> <y> <width> <height>
e.g.
0 0.513672 0.610879 0.246094 0.384937
0 0.567647 0.253906 0.205882 0.156250
1 0.373529 0.246094 0.158824 0.140625
test.txt and train.txt are randomly sampled in 10:90 ratio
format of train.txt and train.txt
imgs/3409.jpg
imgs/894.jpg
imgs/5249.jpg
imgs/1762.jpg
imgs/2337.jpg
imgs/2857.jpg
imgs/5004.jpg
imgs/467.jpg
Download zipped file : https://drive.google.com/file/d/12BAE8t-TumUcOEfJsz4ADL0bka57Dwrw/view?usp=sharing
Credit of images :
https://github.com/X-zhangyang/Real-World-Masked-Face-Dataset