Analysis of Covid-19 Appropriate Mask Wearing Behaviour in Indian Cities Using Deep Learning

It is no secret that Covid-19 is a deadly pandemic. And in a bid to break the chain of transmission, mask-wearing is a widely approved practice outlined by various health experts and the WHO. However, humans often ignore to behave appropriately in response to the pandemic. It has been found that many people do not wear the masks that essentially increases the risk of spreading of the deadly virus in crowded areas. This paper proposes a deep learning-based approach to understand the general behaviour of Indian population across major cities during the onset and continuation of the Covid-19 pandemic. We have proposed a face-mask detection guided method to evaluate the risk factors of various geographical locations (mainly a few important cities of India). Initially, a deep learning-based face mask detection has been proposed to detect the persons without masks, with proper masking behaviour, and with improper masking behaviour. The algorithm takes the image of a public place (e.g. congregated area) as input and detects the human faces in the image with appropriate masking behaviour, We then introduce a new graph-based algorithm to calculate the risk of transmission of the disease based on the mask-wearing behavior of people. In this manner, we intend to keep track of, and analyse, the amount of risk of the spread of Covid-19 over a given time period. Our analysis on several Indian cities show that there are certain links between the daily Covid-19 cases and the mask wearing behaviour.