eye2mouth

train

1. Preprocess cropped images

   $ cd input/
   $ ./untar.sh
   $ cd ..
2. Generate dataset by concatenating and resizing the images

   $ python3 gen_dataset.py
3. Start training

   $ python3 main.py --phase train --dataset_name face128 --batch_size 10 --fine_size 128 --lr 0.00002

test

1. Start testing

   $ python3 main.py --phase test --dataset_name face128 --batch_size 10 --fine_size 128