Face Mask Wearing Detection: A Comparative Analysis

The COVID-19 pandemic has had a tremendous influence around the globe, impacting nearly every element of daily life. It has resulted in widespread illness and death, economic disruption, and changes in societal norms. Governments and organizations have applied various measures to slow the spread of the virus and mitigate its impacts. Among the most important mechanisms is the use of face masks to prevent the transmission and infection of COVID-19. This paper investigates and analyzes different machine learning (ML) methods to execute the classification task of categorizing faces into three classes: wearing masks, not wearing masks, or wearing masks improperly. The preprocessed and augmented dataset used in the study contains 4801 images with the dimension (50, 50, 3) and there are approximately 1500 faces for each class. According to the experimental results, convolutional neural networks (CNNs) can achieve 87% accuracy in classifying faces. These results indicate that CNNs outperform other ML methods, such as random forest, Naïve Bayes, and support vector machine.