eXAI Image Insights

---

BACKEND

To run with Python Virtual Environment

• Clone the project

git clone https://github.com/Error-404-UCD/eXAI

cd eXAI/Backend

• Setup the virtual environment for backend (Discard if pyenv already created)

python -m venv pyenv

• Start the virtual environment

source pyenv/bin/activate

• Install the requirements

pip install -r requirements.txt

• In the backend directory create folders named data and models
• In models directory create another folder checkpoints
• Download and extract the datasets in the data folder such that the path would be example, data/Astronomy/galaxies_heic0007b.jpg
• Make sure that the datasets contain images that store classname_ in it's file name
• Start the server in the Backend folder so that the paths are correctly received in the main file

python app/server_launch.py

• Run unit tests with

python -m unittest discover -s tests

To run with Docker

• Clone the project

git clone https://github.com/Error-404-UCD/eXAI

cd eXAI/Backend

• Build the docker image

 docker build -t docker-exai . 

• Run the docker container. Here the port numbers are in the format > frontend_portnumber:docker_backend_portnumber. Make sure to write the frontend_port number in the React POST request URL (127.0.0.1:5000). HERE frontend_portnumber= 5000 and docker_backend_portnumber= 5000

 docker run -p 5000:5000 docker-exai  

• Wait for the model to train and the server to start. It will display a localhost URL. Note that the port number of this URL is connected to the exposed docker_backend_portnumber.

FRONTEND

• Change Directory to Frontend

cd Frontend

• Install NPM packages

npm install

• Start react server

npm start

---

Download the datasets from here

• MNIST

• BONE MARROW

• ASTRONOMY

---

Refer Google Coding Style guide to push into repository

• Python (https://google.github.io/styleguide/pyguide.html)

• Javascript (https://google.github.io/styleguide/jsguide.html)

---

Dataset sources

• Kaggle Astronomy (https://www.kaggle.com/datasets/subhamshome/esa-hubble-images-3-classes?resource=download)
• Kaggle Bone Marrow (https://www.kaggle.com/datasets/andrewmvd/bone-marrow-cell-classification)
• Kaggle MNIST (https://www.kaggle.com/datasets/scolianni/mnistasjpg)

AWS Setup Guide

• Setup Nginx as reverse proxy on EC2 instance (https://medium.com/@imageadhikari/setup-nginx-as-reverse-proxy-on-ec2-instance-3c6820b6467b)