Skip to content


Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

Creative Adversarial Networks


256x256 samples directly from CAN (no cherry picking) with fixed classification network trained on WikiArt

An implementation of CAN: Creative Adversarial Networks, Generating "Art" by Learning About Styles and Deviating from Style Norms with a variation that improves sample variance and quality significantly.

Repo based on DCGAN-tensorflow.

Getting the Dataset

We used the wikiart dataset available here. Using the dataset is subject to wikiart's terms of use

mkdir data
cd data

Getting pretrained models

We uploaded all of our models to this google drive folder

Training a CAN model from scratch (architecture used in the paper)

bash experiments/ # must run from the root directory of the project

Evaluating an existing CAN model

# make sure that load_dir acts correctly
bash experiments/

External Style Classification network

We ran an experiment where we trained an inception resnet to classify style (60% accuracy) and then used this for the style classification loss, removing the need to learn the layers in the discriminator. We hold the style classification network constant, so the style distribution doesn't change as the generator improves. We found that this improved the quality and diversity of our samples.

Training CAN with External Style Network

# make sure that `style_net_checkpoint` is set correctly, or you will error out
bash experiment/

Training the (ImageNet pre-trained) Inception Resnet

Everything you need should be included in the script. The gist is that it converts the wikiart images into tf records trains the last layer of the model on these images, then fine-tunes the entire model for 100 epochs, at the end of which you should get roughlyy 60% validation accuracy. Since we're looking to generate artwork, this gives us a level of accuracy that is sufficient to try and generate new artwork.

cd slim/
vim # edit INPUT_DATASET_DIR to match the location of where you downloaded wikiart

Evaluating CAN with External Style Network

# make sure that `style_net_checkpoint` and `load_dir` point to the downloaded models. 


We have run a variety of experiments, all of which are available in the experiments/ directory.


Phillip Kravtsov

Phillip Kuznetsov


If you use this implementation in your own work please cite the following

  author = {Phillip Kravtsov and Phillip Kuznetsov},
  title = {Creative Adversarial Networks},
  year = {2017},
  howpublished = {\url{}},
  note = {commit xxxxxxx}