Face Mask Detection Using Convolutional Neural Network

Abstract

The creation of a real-time face mask detection system based on machine learning is the main goal of this research. With the global COVID-19 pandemic, face mask use has become essential for safety and adherence to regulations. The need to identify individuals not complying with the safety measures in public transit, retail settings, and healthcare facilities inspired this project. The goal is to develop a reliable face mask identification algorithm that performs well regardless of the user's mask type, facial angle, or lighting conditions. This research proposes the use of a Convolutional Neural Network based on deep learning, which is trained on a dataset of images of people wearing/not wearing face masks. The TensorFlow framework and Keras API are used to create the model. Transfer learning is also employed by adapting the MobileNetV2 architecture to improve the model's accuracy with less training data. The effectiveness of the proposed model is assessed on two datasets, one containing real-world photos and the other containing artificially generated images. The model performs well, achieving an accuracy rate of 97.5 per cent and 96.8 per cent, respectively, in identifying face masks in real-time. The proposed system has real-world applicability in settings such as hospitals, airports, and other public spaces, where adherence to safety measures is critical. The model can be further improved to detect other types of Personal Protective Equipment such as gloves and face shields. The project concludes with a CNN-based face mask detection algorithm capable of determining in real-time if a person has a mask on or not. The goal is to develop a reliable face mask identification algorithm that performs well regardless of the user's mask type, facial angle, or lighting conditions.