Implementation of Pneumonia Detection using VGG19 based on Chest X-rays

Abstract

Pneumonia is a contagious respiratory disease that results in ulceration and is one of the leading causes of
demise in young children and the ageing people around the world. A Deep Convolutional Neural Network
(DCNN) is proposed in this paper to detect lung inflammation caused by pneumonia. The proposed approach
utilizes X-ray images of chest. A Pneumonia contaminated X-ray images of Chest Data-set. It contains 12,000
pictures of contaminated and non-contaminated chest X-ray pictures are used in the proposed DCNN model.
Image modification techniques and augmentation models are used to create the transformed pictures. The
suggested DCNN application will be created with the VGG19 network and tested on a variety of matrices,
including Accuracy, Precision, and Recall. The agenda of this research is to evolve an authentic and proficient
deep learning model for identifying and classifying pneumonia. To create an end-to-end architecture based
on Convolutional Neural Networks (CNNs), Conditional Adversarial Networks (CAN) are used that maps
ingenuous colors to a grayscale input image. The method, which uses Augmentation techniques, focuses on
increasing generalization in large multi- class picture datasets by enhancing training stability. In limited
domains, augmentation techniques can spawn images that are nearly photo-realistic by using multiple ways,
such as rotating images at 90°, 10% zoom on an image, etc.