Automatic Covid-19 Detection Model from Chest X-ray using CNN, VGG16 and Best Fit Accurate Model

Abstract

This paper describes chest X-Ray images classification like Covid-19 infected chest images or normal chest images using various types of deep learning model. These are VGG16(Transfer learning) and CNN (Convolution Neural Network). Here We made a comparison among VGG16, CNN, SVM models and collected results that which deep learning is more accurate to identify Covid-19 or normal. These models were applied for same dataset and dataset was randomly chosen almost 1350 images (Covid-19 and Normal both) from a website. For this work, at first, we have preprocessed the chest X-Ray image. Then we have extracted the distinct features from the chest X-Ray images. After that, these features have trained into various Deep Learning algorithm and finally classify these images into the category. From the experiment, Convolution Neural Network (CNN) model achieving highest accuracy more than others. The CNN models achieving training accuracy of up to 100% and validation accuracy 94.5% and the VGG16 models achieving training accuracy up to 99.1% and validation accuracy 94.2%. Then validating the CNN model how it detects COVID-19 or normal. After that, best fit accurate model can be easily identified.