this project implements algorithm LAPGAN for high resolution image generation introduced in paper Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks.
install with command
pip3 install -r requirements.txt
train with the following command
python3 train.py --batch_size=<batch size> --num_workers=<workers> --download=(True|False) [--gpus=<gpu number>]
if you train with GPU, --gpus must be over zero.
if you want to train with pretrained checkpoint, run with command
python3 train.py --batch_size=<batch size> --num_workers=<workers> --download=(True|False) --checkpoint=<path/to/ckpt/file> [--gpus=<gpu number>]
save model with the following command
python3 save_model.py --checkpoint=<path/to/ckpt/file>
the saved models will be placed under directory models.