Abstract: Pneumonia kills more children compared to AIDS, measles and malaria combined, pneumonia is detected using chest X-rays diagnosis or also called as Thorax X-rays diagnosis, including the location of the infection and its cause. The following paper discusses the performance of four major transfer learning algorithms such as Mobile Net and Inception V3 on front X-ray images of children of 0 to 5 years of age and classifying the input X-ray. Dataset comprises of 5,000 frontal X-ray images which were augmented to produce 15,000 images to increase the training and testing set. Algorithms are compared with their loss and validation accuracy on how accurately they could classify on a given X-ray image as Normal, Bacterial Pneumonia and Viral Pneumonia after 400 training steps.
Keywords: Pneumonia, TransferLearning, Radiology, Chest, X-Ray.
Title: Diagnosing Child Pneumonia using Transfer Learning
Author: Deep Thakker, Vidit Shah, Jash Rele, Vivek Shah, Dr. Jayashree Khanapuri
International Journal of Interdisciplinary Research and Innovations
ISSN 2348-1218 (print), ISSN 2348-1226 (online)
Research Publish Journals
