Web article

https://nayopu.github.io/eyes2mouth/

How to use

Train

1. Download celeba dataset

   $ ./download_celeba/sh

2. Cropping faces

   $ python gen_dataset.py

3. Start training

   $ python main.py --phase train --dataset_name cropped_128 --batch_size 10 --fine_size 128 --lr 0.00002

Test

1. Start testing

   $ python main.py --phase test --dataset_name cropped_128 --batch_size 10 --fine_size 128