wa8w1r3 / pegans

PEGANs

This is the official implementation of PE-GANs

Requirements

• Python 3.9.0
• Python packages

  # update `pip` for installing tensorboard.
  pip install -U pip setuptools
  pip install -r requirements.txt

Datasets

• CIFAR-10

  Pytorch build-in CIFAR-10 will be downloaded automatically.

• STL-10

  Pytorch build-in STL-10 will be downloaded automatically.

Preprocessing Datasets for FID

Pre-calculated statistics for FID can be downloaded here:

• cifar10.train.npz - Training set of CIFAR10
• cifar10.test.npz - Testing set of CIFAR10
• stl10.unlabeled.48.npz - Unlabeled set of STL10 in resolution 48x48

Folder structure:

./stats
├── cifar10.test.npz
├── cifar10.train.npz
└── stl10.unlabeled.48.npz

NOTE

All the reported values (Inception Score and FID) in our paper are calculated by official implementation instead of our implementation.

Training

• Configuration files

  • We use absl-py to parse, save and reload the command line arguments.

  • All the configuration files can be found in ./config.

  • The compatible configuration list is shown in the following table:

    Script	Configurations
    trainPEGAN.py	PEGAN_P5_CIFAR10_CNN.txt PEGAN_P5_STL10_CNN.txt PEGAN_P10_CIFAR10_CNN.txt PEGAN_P10_STL10_CNN .txt
    trainSNGAN.py	SNGAN_CNN_CIFAR10.txt
    trainWGAN.py	WGAN_CNN_CIFAR10.txt
    trainWGANGP.py	WGAN_GP_CNN_CIFAR10.txt

• Run the training script with the compatible configuration, e.g.,

  • trainPEGAN.py supports training gan on CIFAR10 and STL10, e.g.,

    python trainPEGAN.py \
        --flagfile ./config/PEGAN_P10_CIFAR10_CNN.txt

• Generate images from checkpoints, e.g.,

  --eval: evaluate best checkpoint.

  --save PATH: save the generated images to PATH

  python train.py \
      --flagfile ./logs/PEGAN_P10_CIFAR10_CNN/flagfile.txt \
      --eval \
      --save path/to/generated/images
  

Acknowledgments

Pytorch framework from GNGAN.