Please use this identifier to cite or link to this item: https://gnanaganga.inflibnet.ac.in:8443/jspui/handle/123456789/1099
Title: Early Detection of COVID-19 Using Machine Learning
Authors: Tismeet Singh, Kartikeya Agarwal
Keywords: Computer Vision,
Confusion Matrix
Convolutional Neural Network
COVID-19
Deep Learning
Machine Learing
Transfer Learing
X-Ray
Issue Date: 2022
Publisher: Indian Journal of Computer Science
Abstract: The COVID-19 Pandemic had a devastating impact both on social and economic fronts for a majority of the countries around the world. It spread at an exponential rate and affected millions of people across the globe. The aim of this study was to improve upon a lot of existing studies on COVID detection using Machine Learning. While Machine Learning methods have been widely used in other medical domains, there is now considerable demand for ML-guided diagnostic systems for screening, tracking, analysing, and predicting the spread of COVID-19 and finding a concrete and viable cure for it. We employed the power of Transfer Learning guided Convolutional Networks to predict the existence of the COVID-19 virus in the lung X-Ray of any subject. Deep Learning, one of the most lucrative and potent techniques of machine learning becomes the modern saviour when such crises arise. With the power of this technique, we studied a plethora of models, selected the best ones and then trained them to produce the most optimal results. We used multiple pretrained models and improved upon them by adding structured Dense and Batch Normalisation layers with appropriately selecting activation functions. Elaborate testing yielded a maximum accuracy of over 99%.
URI: http://gnanaganga.inflibnet.ac.in:8080/jspui/handle/123456789/1099
Appears in Collections:Article Archives

Files in This Item:
File Description SizeFormat 
Early Detection of COVID-19 Using Machine Learning.pdf
  Restricted Access
752.04 kBAdobe PDFView/Open Request a copy


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.