Generated never before seen faces from a dataset of Celebrity Faces.
- Implemented a Deep Convolution General Adversarial Network (DCGAN) using Python and Tensorflow.
- Trained the Discriminator and Generator with a loss of 1.09 and 0.76 respectively
Clone the Github repository and use condo to install the dependencies
$ git clone https://github.com/TokyoIndex/dlnd_face_generation.git
$ cd dlnd_face_generation
$ conda install conda
$ jupyter notebook
- Python 3
- TensorFlow 1.0
- Numpy
- Matplotlib 2.0.2
The contents of this repository are covered under the MIT License.