CIFAR-100 GAN

CIFAR-100 GAN is a cutting-edge deep learning solution engineered to generate images from the CIFAR-100 dataset categories. Utilising the prowess of TensorFlow, this model stands as an essential resource for those enthusiastic about delving into generative adversarial networks within the realm of image generation.

Table of Contents

• Features
• Prerequisites
• Installation
• Usage
• Contributing
• Licence
• Acknowledgements

Features

• Train the deep learning model on the CIFAR-100 dataset.
• Monitor and assess GAN convergence and other key performance metrics.
• Employ the advanced TensorFlow deep learning techniques for outstanding image generation.
• Experiment with and adjust GAN architectures.
• Generate images spanning various CIFAR-100 categories.

Prerequisites

• Python 3.7 or higher.
• TensorFlow library.
• A foundational understanding of deep learning, GANs, and neural networks.

Installation

1. Clone this repository:

   git clone https://github.com/amidstdebug/CIFAR-100-GAN.git

2. Navigate to the project directory:

   cd "CIFAR-100-GAN"

3. Install the necessary requirements:

   pip install -r requirements.txt

Usage

1. Ensure you have the necessary data, preferably formatted akin to the CIFAR-100 dataset.

2. Open the IPython notebook to commence the implementation:

   Jupyter Notebook "CIFAR-100-GAN.ipynb"
   

Contributing

1. Fork the project.
2. Create your feature branch (git checkout -b feature/UniqueFeature).
3. Commit your changes (git commit -m 'Add some UniqueFeature').
4. Push to the branch (git push origin feature/UniqueFeature).
5. Open a pull request.

For significant modifications, kindly open an issue first to discuss the proposed changes.

Licence

This project is licensed under the CC BY-NC 4.0 License - please see the LICENCE file for more details.

Acknowledgements

• TensorFlow for providing a powerful foundation for deep learning.
• CIFAR-100 Dataset as the fundamental dataset for this initiative.
• Plus, all those diligent contributors and aficionados who have proffered insights and enhancements to this endeavour!

Should you come across any challenges or have any inquiries, please don't hesitate to raise an issue or contact the project maintainers. We immensely value feedback and contributions!

Back to Top