Recommendation
- Our GAN based work for facial attribute editing - https://github.com/LynnHo/AttGAN-Tensorflow.
Pytorch implementation of WGAN-GP and DRAGAN, both of which use gradient penalty to enhance the training quality. We use DCGAN as the network architecture in all experiments.
WGAN-GP: Improved Training of Wasserstein GANs
DRAGAN: On Convergence and Stability of GANs
left: WGAN-GP 100 epoch, right: DRAGAN 100 epoch
left: WGAN-GP 100 epoch, right: DRAGAN 100 epoch
- pytorch 0.2
- tensorboardX https://github.com/lanpa/tensorboard-pytorch
- python 2.7
You can directly change some configurations such as gpu_id and learning rate etc. in the head of each code.
python train_celeba_wgan_gp.py
python train_celeba_dragan.py
...
If you have installed tensorboard, you can use it to have a look at the loss curves.
tensorboard --logdir=./summaries/celeba_wgan_gp --port=6006
...
- Celeba should be prepared by yourself in ./data/img_align_celeba/img_align_celeba/
- Download the dataset: https://www.dropbox.com/sh/8oqt9vytwxb3s4r/AAB06FXaQRUNtjW9ntaoPGvCa?dl=0
- the above links might be inaccessible, the alternative is