Comparison of Models Architecture on Chest X-Ray Image Classification With Transfer Learning Algorithms

ABSTRACT

Coronavirus disease is an infectious disease caused by the newly discovered coronavirus. Most people infected with the Coronavirus disease have difficulty breathing. Some patients who underwent radiographic tests had changes in their lungs. There are many methods to detect people who have contracted the COVID-19. One alternative method proposed is to perform automatic diagnostics through x-ray images. To perform automation, a model is needed to classify patients with COVID-19 and normal. One of the training methods for creating classifiers is deep learning, but deep learning method needs a huge amount of annotated data. Therefore, the transfer learning method is used by using an existing model, then back again with the desired data. The model used is Xception, InceptionV3 and MobileNetv3. This study produces accuracy of 91% on model trained with Xception, 89% accuracy on model trained with InceptionV3, and 86% accuracy on model trained with MobileNetV3.