Deep Learning for Image Classification with Keras

About The Project

The repository includes Neural Networks implemented using Keras for image classification. Two datasets have been used, the Sign Language Digits Dataset and the Diabetic Retinopathy Arranged Dataset. Both datasets can be found on Kaggle:

• Diabetic Retinopathy Arranged Dataset: https://www.kaggle.com/amanneo/diabetic-retinopathy-resized-arranged
• Sign Language Digits Dataset: https://www.kaggle.com/ardamavi/sign-language-digits-dataset

The project was implemented for the purposes of the Pattern Recognition course of the 9th semester of DUTh.

• More about DUTh and the Pattern Recognition course: https://www.ee.duth.gr/en/course/pattern-recognition/

Getting Started

The notebooks can be found in the Models folder.

Running experiments

There are two notebooks for simple Artificial Neural Networks:

• Notebook ANN_analysis.ipynb contains a series of different ANN architectures.
• Notebook ANN_best.ipynb contains the ANN architecture that performs best.

There are two notebooks for simple Convolutional Neural Networks:

• Notebook CNN_analysis.ipynb contains a series of different CNN architectures.
• Notebook CNN_best.ipynb contains the CNN architecture that performs best.

Contributing

Contributions are what make the open source community such an amazing place to be, learn, inspire, and create.
Contribute following the above steps:

1. Fork the Project
2. Create your Feature Branch (git checkout -b new_branch_name)
3. Commit your Changes (git commit -m 'Add some extra functionality')
4. Push to the Branch (git push origin new_branch_name)
5. Open a Pull Request