/StyleGAN_Image_Detecting

Detecting StyleGAN generated images using CNNs

Primary LanguagePythonMIT LicenseMIT

Detect StyleGAN generated images using CNNs (for Miyazaki-san)

Requirements

pip install -r requirements.txt

Training

Start visdom first：

python -m visdom.server

Then start training via:

# train on gpu0, with no existing checkpoints
python main.py train --load-model-path=None --use-gpu=True

For detail instruction, use:

python main.py help

Testing

python main.py test --data-root=./your_test_dataset --use-gpu=False --batch-size=256