Human Body Temperature Detection based on Thermal Imaging and Screening using YOLO Person Detection

Abstract

In addressing the worldwide Covid-19 outbreak, one of the actions to curb the spread of the virus is to keep people with Covid-19 symptoms away from others in public places. The most easily detected symptom is high body temperature due to fever, and this is one of the common symptoms of Covid-19 patients. Several methods of body temperature detection have been implemented at the entrance of the premises. One of the most common methods is to use an infrared temperature scanner. This method has some constraints including its use which is time-consuming and can lead to further spread of the virus as gun-type scanners can be a medium of virus spread as has been held by many people. Another more advanced method is the detection of body temperature through a thermal camera with imaging. Although more sophisticated, this method also has the constraint where the temperature is usually detected as a whole and does not differentiate the temperature of the human body and other nearby objects. With a focus on this problem, this study applies a combination of object detection methods through image processing with temperature detection through thermal imaging. For the object detection process, the You Only Look Once (YOLO) model and the OpenCV library have been used, especially in detecting people and non-people. While the calculation of body temperature through thermal images has been made more accurate because the scanned temperature is more specific based on the detected objects. In this way, a person’s body temperature can be separated and will not be affected by the temperature of other objects. From the results and analysis obtained, an accuracy of 100% can be achieved based on a pre-trained model for human body temperature detection. With more specific and accurate detection as produced in this study, then a warning or caution will be issued only when a person actually has a high body temperature and will then not be allowed to enter the premises.