SNGAN

An implement of spectral normalization GAN for tensorflow version, spectral normalization is used in this repo to give some constrants to the weights of discriminator

Requirements

(1)download this repo to your own directory

$ git clone https://github.com/nnUyi/SNGAN.git
$ cd SNGAN

(2)download celebA dataset and store it in the data directory(directory named data)

celebA datasets is cropping into 64*64 size with *.png or *jpg format, this repo read image format data as input.

(3)training

$ python main_sngan.py --is_training=True

(4)testing

Anyway, in this repo, testing processing is taken during training, It samples the training results to the sample(named sample) directory, and stores session in the checkpoint(named checkpoint) directory.

sample image	sample image

64*64 resolution	64*64 rosolution

Name		Name	Last commit message	Last commit date
Latest commit History 12 Commits
data		data
README.md		README.md
SNGAN.py		SNGAN.py
main_sngan.py		main_sngan.py
ops.py		ops.py
utils.py		utils.py