Detecting Face Masks and Social Distancing Against COVID-19 with Embedded Systems and Deep Learning Technologies

Social distancing and face mask wearing have been proven as effective measures against the spread of the infectious COronaVIrus Disease 2019 (COVID-19). However, individuals are still adapting to COVID-19 regulations. In fact, you can often see people in public places wearing face masks incorrectly or not wearing face masks at all, besides not tracking the required two meters (6 feet) distance between themselves and their surroundings. An active surveillance system that can both determine whether or not a person is wearing a face mask and tracking distances between individuals would be able to correctly report when they are at risk of being exposed to the disease. Thus, such active surveillance system would be able to keep people safer and slow down the spread of the disease. On the other hand, recording data and labeling individuals who do not follow the measures would breach individuals' rights in free societies. Thus, in this work we present a deep learning-based real-time social distancing monitoring and face mask detection system considering two important ethical factors: (i) the system should never record/cache data, and no human supervisor should be in the detection loop. Given the limited computational capabilities of embedded devices, achieving near real-time processing times while still retaining high accuracy in terms of both object localization and classification is an extremely challenging task. To this end, a trade-off between inference speed and detection accuracy must be found.