Mask Detection From Face Images Using Deep Learning and Transfer Learning
Ahmet Haydar Örnek, Mustafa Çelik, Murat Ceylan · 2021
It is vital important for people to wear masks during the pandemic that affects the whole world. In this study, it was detected whether people wear masks by using convolutional neural networks which is one of the deep learning methods and transfer learning. In the classification carried out using the Resnet-18 architecture, both real-life images obtained with the Huawei M2150 camera and images shared as open source were used. The system, which was trained using 18600 images, was tested with 4540 real-world images and 95.16% sensitivity 96.69% specificity values were obtained. Thus, a model that works with high performance not only on high resolution images taken close up, but also on low resolution images taken from afar was obtained.