Skip to content

Python implementation of the Deep Convolutional Generative Adversarial Network (DCGAN) for generating realistic fake faces using PyTorch.

License

Notifications You must be signed in to change notification settings

Pinak-Datta/Fake-Face-Generation-Using-GAN

Repository files navigation

Fake-Face-Generation-Using-GAN

License: MIT

This project is a Python implementation of the Deep Convolutional Generative Adversarial Network (DCGAN) for generating realistic fake faces using PyTorch. The network architecture used in this project is inspired by the paper "Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks" by Alec Radford, Luke Metz, and Soumith Chintala.

Dependencies

Python 3.6+
PyTorch 1.9+
TorchVision 0.10+
Matplotlib

Results

Output for 1000 Epochs Another Generated Output for 1000 Epochs

Installation

  1. Clone the repository
git clone https://github.com/Pinak-Datta/Fake-Face-Generation-Using-GAN.git
  1. Install the dependencies using pip:
pip install -r requirements.txt

Usage

To generate fake faces using the pre-trained model, run the following command: (Note that here you need to supply the path to the image you want to create a fake of)

python generate.py

To train the model on your own dataset, place your images in the data directory. To increase and decrease the number of epochs, change the num_epochs and run the following command:

python Train.py

Acknowledgements

This project is inspired by the work of Alec Radford, Luke Metz, and Soumith Chintala on DCGAN. The pre-trained model used in this project is based on the CelebA dataset.

About

Python implementation of the Deep Convolutional Generative Adversarial Network (DCGAN) for generating realistic fake faces using PyTorch.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages