DeepClassifyML

A Web Application made with Flask to classify Respiratory Diseases from Chest X-rays

Requirements

• Python 3.6
• Google Collab (To re-train or fine tune the model)

Installation

Web App

1. Clone the repository and open DeepClassifyML
2. Use pip install -r requirements.txt to install all the packages necessary to run the Web App.

Jupyter Notebook

1. Export the file Inception_v3_fine_tune.ipynb to Google Colab if you want to fine tune the model and export your own model.h5 file.

Running it

There are currently two ways of running this and obtaining predictions.

1. URL Prediction

To use this option run python3 server.py and open 0.0.0.0:8080 in your browser. This will let you add a chest x-ray image already uploaded to the internet and diagnose the the patients condition.

2. Image Upload Prediction

To use this option (which is still very buggy at the momemnt) run python3 server_new.py and open 0.0.0.0:8080 This will let you upload a chest x-ray image file from your system and diagnose the the patients condition.

Notes

• The Dataset used to train the model is ChestXray14 you can find out more about it here
• Transfer Learning was used to make ImageNet predict and give output pased off of the dataset
• The Image Upload method is currenty very buggy, WIP.

Contributors

• Prince Canuma
• Daniel Phiri
• Vishwa Mehta
• Mayur Patil
• Parth Purohit

Based off this Image Classifer Web Application. Feel free to open up an issue or contribute to the project.