NOTE: The setup and example code in this README are for training GANs on single GPU. The models are smaller than the ones used in the papers. Please go to link if you are looking for how to reproduce the results in the papers.
Official Chainer implementation for conditional image generation on ILSVRC2012 dataset (ImageNet) with spectral normalization and projection discrimiantor.
Consecutive category morphing movies:
pip install -r requirements.txt
Please download ILSVRC2012 dataset from http://image-net.org/download-images
cd datasets
IMAGENET_TRAIN_DIR=/path/to/imagenet/train/ # path to the parent directory of category directories named "n0*******".
PREPROCESSED_DATA_DIR=/path/to/save_dir/
bash preprocess.sh $IMAGENET_TRAIN_DIR $PREPROCESSED_DATA_DIR
# Make the list of image-label pairs for all images (1000 categories, 1281167 images).
python imagenet.py $PREPROCESSED_DATA_DIR
# Make the list of image-label pairs for dog and cat images (143 categories, 180373 images).
python imagenet_dog_and_cat.py $PREPROCESSED_DATA_DIR
python source/inception/download.py --outfile=datasets/inception_model
LOGDIR=/path/to/logdir
CONFIG=configs/sn_projection_dog_and_cat_64.yml
python train.py --config=$CONFIG --results_dir=$LOGDIR --data_dir=$PREPROCESSED_DATA_DIR
LOGDIR=/path/to/logdir
CONFIG=configs/sn_projection_64.yml
python train.py --config=$CONFIG --results_dir=$LOGDIR --data_dir=$PREPROCESSED_DATA_DIR
(If you want to use pretrained models for the image generation, please download the model from link and set the snapshot
argument to the path to the downloaded pretrained model file (.npz).)
python evaluations/gen_images.py --config=$CONFIG --snapshot=${LOGDIR}/ResNetGenerator_<iterations>.npz --results_dir=${LOGDIR}/gen_images
Regarding the index-category correspondence, please see 1K ImageNet or 143 dog and cat ImageNet.
python evaluations/gen_interpolated_images.py --n_zs=10 --n_intp=10 --classes $CATEGORY1 $CATEGORY2 --config=$CONFIG --snapshot=${LOGDIR}/ResNetGenerator_<iterations>.npz --results_dir=${LOGDIR}/gen_morphing_images
python evaluations/calc_inception_score.py --config=$CONFIG --snapshot=${LOGDIR}/ResNetGenerator_<iterations>.npz --results_dir=${LOGDIR}/inception_score --splits=10 --tf