A Deep Learning Fusion Approach for Mask Detection: CNN and VGG16 Integration

Abstract

Since the COVID-19 epidemic began, wearing a face mask has become a crucial precaution to stop the virus from spreading. In this research, we present a mask detection system that combines the VGG16 model with a Convolutional Neural Network (CNN) architecture. To do this, we construct a comprehensive dataset image of people with and without masks set against diverse backgrounds. The training, validation, and testing sets are then created from the dataset. The pre-trained VGG16 model is utilized as a feature extractor to pull out distinctive qualities from the input images. The outcomes show how well the fusion of CNN with VGG16 model can distinguish between masked and unmasked people even in difficult situations involving occlusions and a wide range of backdrops. We demonstrate the proposed method's better accuracy and computational effectiveness by comparing it to state-of-the-art mask detection approaches. The system is a trustworthy tool for mask detection in situations in the real-world including airports, hospitals, and public areas since it achieves an overall accuracy of over 99.47% of training and 98.13% of validation respectively.