watsonyanghx / gan_lib_tensorflow

Tensorflow implemention of various GAN.

GAN_Lib_Tensorflow

Tensorflow implemention of various GAN.

Please refer to corresponding folder for more details.

Prerequisite

• Python 3.5.4
• TensorFlow 1.5
• Numpy
• Scipy

Quantitative evaluation

Method	Inception (This repo)	Inception (Official)	FID
Real data	12.0	11.24	3.2 (train vs test)
PGGAN	-	8.80 ± 0.05 (-, Unsupervised)	-
SNGAN	8.43 ± 0.12 (ResNet, Supervised)	8.24 ± 0.08 (ResNet, Unsupervised)	-
ACGAN	7.86 ± 0.09 (ResNet, Supervised)	8.25 ± 0.07 (-, Supervised)	-

Inception scores are calculated by average of 10 evaluation with 5000 samples.

TODO

• FID

• MS-SSIM

Generated images

• SNGAN

  

• ACGAN

  

Reference

• https://wiseodd.github.io/

gan_lib_tensorflow's Issues

How to use in 3D conv?

The paper and the code are both for the 2D convolution of the sn limit w, then how to deal with w in the 3D convolution?

Projection discriminator

Hello,

Do you have an idea of how to use Embed_Id layer in tensorflow for projection discriminator?

Thanks,
Ansh