Deep Learning Framework for Face Mask Detection

Jeny, J. R. V.; Shraddha, B.; Ashritha, B.; Sai, D. S.; Naveen, M.

COVID-19, a pandemic disease spread caused by the corona virus, has taken a toll on our lives. Wearing face masks, sanitizing, and maintaining social distance has been the new normal since the outbreak. As the Corona virus enters our bodies through the mouth and nose, covering those areas can help us to protect ourselves from the virus to some extent. Although vaccination and medications are issued for this disease, WHO states that wearing masks and following Covid protocols should be continued. After the first wave has reached to a control and vaccination started, public has started neglecting face mask and social distancing. This resulted in a massive spread of the second wave, over which some populated countries are losing control. Henceforth, wearing a mask will highly help in preventing the transmission of the disease. This project helps the detection of faces with and without mask from live video streams. By making use of packages like TensorFlow, SciPy, OpenCV, Keras the implementation is done. The model used for face detection is MTCNN. The overall idea of developing this project is to implement an embedded model, which can detect faces from video streams and generate a valid classification of faces with/without mask. This model will be useful for the public surveillances and insist people to put on their masks and prevent the spread of dangerous corona virus. The proposedmodel has obtained a classification accuracy of 97.2%. © 2021 IEEE.