Deep learning framework for smart patient attendant system using thermal images

• Main Authors: Dusa, Dinesh, Gundavarapu, Mallikarjuna Rao
• Format: Conference Proceeding
• Language: English
• Subjects: Algorithms; Artificial intelligence; Deep learning; Disease control; Face recognition; Facial recognition technology; Infectious diseases; Infrared imagery; Infrared imaging; Infrared signatures; Machine learning; Thermal imaging; Viral diseases
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0187049

A terrible disease, Covid-19, has afflicted the world. As it spreads infectious and air droplet transmission disease, it is always mandatory to wear masks and temperature checks in offices and crowded environments. As a matter of fact, the temperature check has become more mandatory. The manual way of measuring temperature also may spread the disease. A process that is fully automatic and does not require human interaction has a considerable advantage in controlling the spread. A camera with an IR sensor equipped with artificial intelligence can help with this. In epidemics involving infectious disease pandemics such as Ebola and SARS, using infrared thermographs (IRTs) as a fever detection method has been implemented. As Infrared images show the unique heat-signature of the human face, which can be used for facial recognition also which provides an added advantage. IR images have inherent advantages over visible light images due to their characteristics, making them useful in various fields such as improving face recognition algorithms. It is evident that IR images remain constant even in extreme lighting conditions. As well as detecting facial temperature based on pixel intensity, this paper deals with the deep learning based facial recognition and temperature detection from facial thermal image dataset. We are proposing a CNN based densenet201 architecture for facial recognition and identification, and OpenCV methodologies for examining the temperature of the validated images to accomplish this experiment.