Face Mask Detection and Social Distance Monitoring with Deep Learning
Aravind Sharma Kala, P. Sharon Femi, V. Rajalakshmi, S Kalavathi, Ashwini K M · Indian Journal of Science and Technology · 2023
Objectives: This work proposes a real-time classification model that can accurately detect whether an individual is wearing a face mask and maintaining social distance with the goal of developing a lightweight and easily deployable system for surveillance purposes. Methods: The proposed method easily identifies the human by bounding boxes and wearing of face mask by realtime Face Detection Recognition System. This is a robust model that involves detection, tracking and validation as its features. Pre trained deep learning models like Inceptionv3, DenseNet are used and compared with the proposed Improved MobileNetv2. A tested deep learning model is developed to check social distancing, which uses the YOLO object detector and computes the Euclidean distance between people to confirm the safety of the system. Finally, the proposed method is evaluated in terms of precision, recall, F1-score, support, accuracy, sensitivity and specificity. Findings: The analysis of the results reveals that the improved MobileNetv2 model achieves precision of 98%, recall of 98% and accuracy of 98%. Hence, this deep learning system contributes to the management of COVID-19 outbreak in an efficient way and can be installed for operation in public places like shopping malls, stadiums etc. Novelty: The proposed system can be effortlessly integrated into embedded devices that have limited computational capabilities to the detection of face masks in photographs and real-time videos. Keywords: Human Detection; Improved Mobilenetv2; Face Mask Detection; YOLO