Dataset for Face-mask Recognition in Poor Visibility Conditions based upon IoT enabled Robotics

Abstract

There has been a significant increase in demand and use of facemask after the increasing transmission of the Corona Virus. Wearing face masks can help in reducing the spread of the virus from one person to another. But some people still don’t wear a mask and checking it manually in a huge crowd can be very difficult and tedious. Various face mask detection systems have been made for making this task easy. In poor visibility conditions detecting facemasks becomes more difficult. The ubiquity of haze substantially reduces the quality of images. To restore the quality of hazy image various image dehazing algorithms have been designed by researchers. However, there are not many studies that encapsulate dehazing algorithms and techniques used for spotting objects (here, facemasks) based on deep learning. This paper aims to propose an idea for spotting face masks in extremely poor visibility conditions by creating a dataset of images captured in different densities of haze by a robot using digital image sensors.