Unofficial implementation of Swapping Autoencoder for Deep Image Manipulation (https://arxiv.org/abs/2007.00653) in PyTorch
First create lmdb datasets:
python prepare_data.py --out LMDB_PATH --n_worker N_WORKER --size SIZE1,SIZE2,SIZE3,... DATASET_PATH
This will convert images to jpeg and pre-resizes it. This implementation does not use progressive growing, but you can create multiple resolution datasets using size arguments with comma separated lists, for the cases that you want to try another resolutions later.
Then you can train model in distributed settings
python -m torch.distributed.launch --nproc_per_node=N_GPU --master_port=PORT train.py --batch BATCH_SIZE LMDB_PATH
train.py supports Weights & Biases logging. If you want to use it, add --wandb arguments to the script.
You can test trained model using generate.py
python generate.py --ckpt [CHECKPOINT PATH] IMG1 IMG2 IMG3 ...
