A generative adversarial network that creates new Pokémon from a random noise vector.
- Raw images are located under the
raw_dataset
directory - To create
training_dataset
, runpython image_preprocessing.py
- Open
PokemonGAN.ipynb
in Colab and follow instructions in the notebook. - Root folder:
My Drive/PokemonGAN
- Training samples generated by
vis_noise.npy
will save toPokemonGAN/training_samples
. - Weights will save to
PokemonGAN/weights
- Once training is complete, download the
training_samples
directory from your Drive - Run
python make_training_GIF.py