Monitoring COVID-19 prevention measures on CCTV cameras using Deep Learning

With the unfortunate rise of COVID-19 disease the whole world is in search of a way to stop the spreading of the virus. New restrictions in everyday life came up, like quarantine, social distancing, wearing masks, washing hands more frequently, limited number of people in closed places and more. This project is dedicated to monitor, through CCTV cameras videos, three of major prevention measures: social distancing, wearing masks, counting the number of people in a closed place. Besides being three different problems distant from each others, they share the possibility of being analyzed through camera images. The need of having some indicators in real time that these measures are being respected in an area of interest is more current than ever. Moreover, in these days CCTV cameras can be found anywhere, from public places such as airports, hospitals, schools, museums to shops, retail stores, houses. It makes the perfect instrument to reliably have real time images with no further installations. With the improvements in the last years in the field of GPU computing, Machine Learning algorithms became the first candidates to address this kind of problems. Different kinds of models in this work are used and presented to offer a reliable instrument to counteract the spreading of the virus.