Exploring the Performance of Various Classification Algorithms for Masked Face Recognition Using CNN-Based Features

The COVID-19 causes global pandemic which affects human health. As per the report, COVID-19 patients were suffering from reasonable symptoms and recuperate without further care. Many people who are affected need proper medical support. For the effective protection, the best method is using a face mask. Here in this paper, we are trying to predict the best algorithm for a masked face. Fig. 1 shows the blueprint of how the system works. This model proposes two components. Feature extraction using CNN is used as the first component and identifies the masked and non-masked face, and random forest, SVM, Naïve Bayes, decision tree and KNN are used for the classification process. Exponential findings suggest that random forest produces more accurate results on the dataset we developed, which consists of roughly 80 photos with 98.24% accuracy.