Nasrul, Andi Baso Kaswar, Satria Gunawan Zain, Fhatiah Adiba, Dyah Darma Andayani, Andi Akram Nur Risal
Pneumonia is a disease in which a person experiences an infection of the lower respiratory tract along with symptoms such as coughing and shortness of breath. The main causes of pneumonia are microorganisms such as viruses, bacteria and fungi. This study aims to compare the performance of Convolution Neural Network CNN) architecture in detecting pneumonia. There are three architectures compared, namely MobileNet, ResNet, and DenseNet. In this study, 4440 x-ray images of human lungs were used. Based on the tests, MobileNet has an accuracy of 97% with a computation time of 9.55 seconds, ResNet has an accuracy of 95% with a computation time of 36.81 seconds, and DenseNet has an accuracy of 95% with a computation time of 26.23 seconds. Based on the results of this study, it shows that the MobileNet architecture has a better performance in detecting pneumonia disease compared to the other two architectures. © 2024 IEEE.
State University of Makassar, Departement of Computer Engineering, Makassar, Indonesia
