Real Time Social Distancing and Mask Detection using Convolutional Neural Networks

The coronavirus disease officially named as COVID-19 pandemic has brought global crisis with its deadly spread to more than 180 countries. Though vaccines are introduced, it is essential for everyone to follow certain precautionary measures. This paper presents the methodology for social distancing and mask detection using computer vision and machine learning technique to reduce the spread of coronavirus pandemic. Real time inputs are provided to the system through computer webcam. Image processing steps such as video to frame conversion, preprocessing and feature extraction are performed. The model for mask detection is trained on a dataset that consists of images of people with and without masks collected from various sources using MobileNetV2 which is very efficient for object detection and segmentation. Faces are detected using Caffe model and trained model is used to detect the presence of mask along with its probability. The Euclidean distance between the faces is calculated and an alert is generated if the distance is below the threshold value. It is hoped that our study would be useful to mitigate the impact of this communicable disease for many countries in the world.