Tensorflow implementation of Globally and Locally Consistent Image Completion on celebA dataset.

• smaller image input size (128x128)
• smaller patch sizes
• less number of training iteration (500,000 iterations in the paper)
• Adam optimizer used instead of Adadelta

• Opencv 2.4
• Tensorflow 1.4

-data
  -img_align_celeba
    -img1.jpg
    -img2.jpg
    -...

$ python train.py 

To continue training

$ python train.py --continue_training=True

$ python test.py --img_path=./data/test/test_img.jpg

Use your mouse to erase pixels in the image.
When you're done, press ENTER.
Result will be shown in few seconds.