DCGAN-Tensorflow

Implementation of Deep Convolutional Generative Adverserial Network (DCGAN) in Tensorflow with support for Tensorboard and configured to be trained on CelebA dataset.

Generator

Disciminator

Requirements

Here are the requirements to run DCGAN-Tensorflow

• Python 2.7 or Python 3.3+
• Tensorflow >= 0.12
• Numpy
• tqdm
• json
• Bunch

Usage

Edit the values in config.json file to fit your experiment as follows:

{
    "exp_name" : "dcgan_small_64",
    "num_epochs": 20,
    "batch_size": 128,
    "img_size":64,
    "lr":0.0002,
    "beta1":0.5,
    "summaries_period":100, 
    "gf_dim":64,
    "df_dim":64,
    "data_dir": "../img_align_celeba",
    "noise_shape":100,
    "max_to_keep":3
}

summaries_period: How frequently summaries are written for example 20, then it will write summaries after every 20 steps.

gf_dim: Number of filters in the last layer in the Generator network.

gf_dim: Number of filters in the first layer in the Discriminator network.

max_to_keep: Maximum number of checkpoints kept in the checkpoints directory.

And then you run the code as follows:

python main.py

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests as appropriate.

License

MIT