A hybrid machine learning and fuzzy inference approach with UAV for indoor virus contamination risk

Abstract

With the impact of the Covid-19 pandemic in 2020, major established health rituals were forced to transform. The most well-known of these is the medical mask, which is widely used and required to be worn in designated areas. Although pandemic regulations have been relaxed recently, health authorities agree that wearing masks, especially in closed areas, is a life-saving measure. Proper use of face masks is one of the most effective, easy and inexpensive actions to prevent the rapid spread of viruses indoors. By examining the use of masks in closed areas, the risk of transmission of the virus can be analyzed, and the measures can be determined correctly. Taking advantage of up-to-date technological equipment and approaches are important tools for making these determinations accurately and easily. In this study, the risk of indoor virus transmission from mask wearing styles is analyzed with an integrated method that includes Machine Learning (ML) and Fuzzy Inference System (FIS) approach. In order to achieve this, images taken from the camera of the Unmanned Aerial Vehicle (UAV), which is one of the current technologies suitable for contactless, mobile operations, were used. While determining the mask wearing status with the help of machine learning over the images, the ambient temperature and the mask wearing ratio gave the risk results with the fuzzy inference system. The results are intended to guide decision makers in identifying and implementing measures to reduce and prevent the spread of the virus indoors.