ABSTRACT
The antigen test kits or ATKs have been widely used for screening COVID-19 infections because they can detect and give the results quickly and can be done easily by untrained patients. However, reading ATK test results could be difficult for some people and may lead to misinterpretations of the test results. This paper presents a preliminary study for developing a mobile application for helping in reading the results of the COVID-19 ATKs from an image using algorithms based on the YOLO object detection. The results are classified into 3 classes, negative, positive, and invalid. The negative and the invalid results are further refined by using the distances between the visible line and the letters on the test cassette. Experiments were conducted to test the efficiency and accuracy of the developed model with a mean of average precision or mAP of 0.986 and an F1 score of 0.970. The model was developed and put into a prototype mobile application using tools that support cross-platform technology. © 2022 Asia-Pacific of Signal and Information Processing Association (APSIPA).