This is created for wearing mask detection. We have our paper "Hybrid Transfer Learning and Broad Learning System for Wearing Mask Detection In the COVID-19 Era" has been accepted by "IEEE Transactions on Instrumentation & Measurement" with DOI 10.1109/TIM.2021.3069844.
Dataset:
1、------WMD------ We create a dataset named wearing mask detection (WMD). It can be downloaded by: https://pan.baidu.com/s/1gfHD821Kej1ZSyM1BV-wLQ extract code：0gyw .
It includes 7804 images with a total number of 26403 wearing masks. It is divided into three parts:
Training set: 5410 images, with 17654 wearing masks
Validation set: 800 images, with 1936 wearing masks
Testing set: 1594 images, with 6813 wearing masks.
2、-----WMC-------
Wearing Mask Classification Dataset can be downloaded by： https://pan.baidu.com/s/1tLtt6CkHDj5ACI7g7EVi8Q extract code: jb7y . It includes 19590 mask samples and 18555 background samples.
3、--------Face dataset-----
It includes 4054 images with a total number of 16216 faces.
Traiing set: 2407 images with 9368 faces
Validation set: 500 images with 1798 faces
Testing set: 1147 images with 5032 faces
They can be downloaded by https://pan.baidu.com/s/1jtKyGP4KMzKKIvrRg-PJQg extract code: 86av .
Running environments:
*Windows 10 OS *Tensorflow 1.5 *NVIDIA Geforce GTX 1660 Super with 6 GB memory
Thanks EdjeElectronics' work about how to use TensorFlow for object detection on Windows. https://github.com/EdjeElectronics/TensorFlow-Object-Detection-API-Tutorial-Train-Multiple-Objects-Windows-10#2-set-up-tensorflow-directory-and-anaconda-virtual-environment
If you use this dataset or for a comparison, please cite it.
@article{wang2021hybrid, title={Hybrid Transfer Learning and Broad Learning System for Wearing Mask Detection In the COVID-19 Era}, author={Wang, Bingshu and Zhao, Yong and Chen, CL Philip}, journal={IEEE Transactions on Instrumentation and Measurement}, year={2021}, publisher={IEEE} }
Notably, the dataset is only used for academic use. If you want to use it for commercial use, please contact us by [email protected] .
We show special thanks to those who help us. Detailed acknowledge is presented in our paper. Hopefully our work can provide some help in the fighting against COVID-19.