A Research For Face Mask Recognition Based On The CNN Model

Abstract

Face recognition has become popular in the world today. This technique can be applied in the following fields: class attendance, authenticating users, and keeping track of people in a large space. Due to the outbreak of COVID-19, people all over the world are forced to maintain social distance. There is a requirement to wear a mask in public and keep safety for everyone, like washing hands or notifying the authorities about the symptoms. However, the practical application of the old model is limited. Because of the lack of ability when recognizing face with masks, people must take off their mask when using these systems. This increases the risk of infection. The aim of this thesis is to research the state-of-the-art methods for face recognition and existing applications in such field. After that, there is a proposed solution to recognize both masked and non-masked faces. The proposed method consists of the following steps: (I) a training dataset is created using the existing dataset and image processing technique, (II) train model with new dataset, (III) implementation of face recognition application based on new model. Comparing to several methods, our method has better result when recognizing face with mask. When testing on non-masked face dataset, my model declined in accuracy.