A variational autoencoder that generates images of cats.
The cat model was trained on cropped versions of the data from the Kaggle cat dataset obtained from this git repository. The training data consistes of 9997 100 * 100
pixel images of cat faces.
The model can be trained by running trainModel.ipynb
with appropriate parameters set (path to the image data). The weights that result from training the model with default parameters are available in the ./weights/
directory.
The model may be tested by running testModel.ipynb
.
Individual cat images can be generated using the provied script genCats.py
. The script can be run using
python3 genCats.py <desired number of cats>