In this project, I have built and trained a DCGAN on a dataset of human faces. The goal of this project is to get a generator network to generate new images of faces that look as realistic as possble. Check the images at the end of the notebook to see the final results
-
Clone the repository and navigage to the downloaded folder.
git clone https://github.com/eswar3/face-generation-using-GAN.git cd face-generation-using-GAN
-
Open the
face_generation.ipynb
file. Of course, you can find HTML version of the file.jupyter notebook face_generation.ipynb
-
Read and follow the instructions in the jupyter notebook.
-
This repository does not include the dataset. You can find the instruction in notebook on how to downlaod it.
- Pre-processed Data
- Define the Model
- Discriminator
- Generator
- Build complete network
- Discriminator and Generator Losses
- Optimizers
- Training
- Generator samples from training
- tensorflow (Generator and Discriminator)
For training the model I recommend you to use a GPU.
You can use Amazon Web Services to launch an EC2 GPU instance. (This costs money!)