keras-cnn

A CNN for classification of retinal fundus images built using Keras with tensorflow backend

Project Structure

Directory Tree

./
├── data
│   ├── test
│   │   ├── 0
│   │   ├── 1
│   │   ├── 2
│   │   └── 3
│   └── train
│       ├── 0
│       ├── 1
│       ├── 2
│       └── 3
├── environment.yml
├── model
│   ├── model.h5
│   ├── model.json
│   ├── model.png
│   └── model.txt
├── result.html
└── src
    ├── export.py
    ├── model
    │   ├── cnn.py
    │   ├── __init__.py
    │   └── __pycache__
    ├── test.py
    └── train.py

To run the project,do:

1. Install miniconda or anaconda
2. Open the project folder in terminal
3. Load the environment.yml file using the command below

conda env create -f environment.yml

4. Activate the environment in conda
5. Run the src/train.py to train the model

python src/train.py

6. Export the neural network model to text,image and json using the src/export.py file

python src/export.py

7. Test the neural network using external images with src/test.py

python src/test.py