Attribute2Sketch2Face
This repository is for our work on Face Synthesis from Visual Attributes via Sketch using Conditional VAEs and GANs.
Dependencies:
python 2.7
Pytorch
Data:
(1) please download the preprocessed [CelebA] and LFWA database by link:
(2) extract them into
Training:
-
change the command direction into
$root$ -
when training on the LFWA dataset. please run: python ./train_lfw.py
-
when training on the CelebA dataset. please run: python ./train_celeba.py
Testing:
-
when testingon the LFWA dataset. please run: python ./test_lfw.py
-
when testing on the CelebA dataset. please run: python ./test_celeba.py
-
All the test result should be in the directory:
$root$ /result/
Acknowledgement:
This code is highly inspired by the following:
StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networks:
Generative Adversarial Text-to-Image Synthesis:
Also, we highly appreciate the help from He Zhang on the Densely Generator "G2" in this code.