The problem Brain Tumor Classifier Web-App solves

During this Pandemic we have realised the importance of good health care facilities and as AI is already revolutionizing the healthcare department, we decided to build a web app that would help in classifying Brain tumor through MRI scans.
The main objective behind making this project was to attempt to make these sorts of health care facilities accesible to all. As it is a web app, all it needs is an internet connection to run. It reduces the amount of time a person has to visit a clinic or hospital and with such a tool available to general public, it would help in reducing the expenditures on healthcare facilities.

It focuses majorly on supporting those middle class and below poverty line families who won't be able to afford these high expenditures and i think this idea can be highly beneficial for them. It also reduces the cost of equipments and other things related to reading and detecting these brain tumors for the hospitals and clinical facilities. As the project is open source, students can use it to learn about how the models work and can also contribute in enhancing the overall product.

Challenges we ran into

Here are a few challenges that faced while building the project:-

1. The model which was trained initially didn't give quite satisfactory results as the rate of mis-classification was high. We later on improved the model by adding more convolutional and maxpooling layers and also tried to prevent overfitting by methods such as using dropout layers.

2. Our main focus was deploying the model and we started off by using Flask, but we ran into lots of troubles as we were not able to connect the model correctly with the web page. We chose to go with Stream lit and Ngrok which were much more helpful in deploying our model.

3. We weren't able to launch our server initially but after going through the streamlit documentations , we were able to find a way to run our server.