Covid Helper
It is a website that can detect the Covid-19 through x rays using deep learning. It shows nearest vaccination centre also it provides resources for fighting covid and getting vaccinated .
Created on 2nd July 2021
•
Covid Helper
It is a website that can detect the Covid-19 through x rays using deep learning. It shows nearest vaccination centre also it provides resources for fighting covid and getting vaccinated .
The problem Covid Helper solves
As you know our Project is directly related to present pandemic going worldwide i.e Covid-19 . So inorder to help society to fight against it , we had developed and website using deep learning that will give result as soon as he/she insert XRAY and can successful knows whether he/she is Covid-19 or he is Normal. This in turn will help prevent propagation of virus because a positive patient will restrict his movement as soon as he gets test results, and through this method test results will be obtained faster
So it has many benefits to our society ,
a. Saving Time : It can save plenty of time of peoples than simpling going in LAB and waiting for 2-3 days for coming out result. Also it Can process multiple results at a time , Also website has very good interface , so it helps user to access or work on it smoothly .
b. Mental Health : As we know for actual test of covid-19 there is very difficult process for testing , as they take sample from Nose etc , so many people fear for doing test and indeed it affect there mental health.
c. Cost efficient : These is a simple CNN model and a website , so User just had to upload it’s X-ray , so we can look there is very less expense.
Challenges we ran into
While selecting the Model , we were having two options Resnet 18 or Resnet 50 . So we selected Resnet 50 , because it was having more accuracy as compared to Resnet 18 .
Dataset which was taken for training , was containing very large number of X-rays , so there was problem of Dataset Sizing i.e Problem of Overfitting. To overcome that , we Used Modified randomized oversampling technique .
Before designing of website we decided to make just application of python , but we did not find that suitable method to express our project , so decided to make website for it . So there was Deployment Issues i.e Integrating Resnet 50 model with Django . So we used Keras library to make it effective .
When we were testing our model , accuracy was very low which was not effective . So our aim was to create a model with high Accuracy Issues i.e Improvise and Optimization to maximum . So we first used Resnet 18 and then used Resnet 50.
As we first decided to make just python application , but it was not user friendly . So we decided to make website .Creating User friendly UI.
Technologies used