SphereGAN-Pytorch-implementation

Pytorch implementation of SphereGAN(Sung Woo Park and Junseok Kwon)

cf) The Generator and Discriminator structures is not the same in the paper. In this repository, the Generator and Discriminator structures are used DCGAN’s structures.

Requirments

Code is written in Python 3.7.3(Pytorch 1.1.0) and requires:

• Pytorch
• tqdm

Run the example

MNIST, CIFAR-10

python main.py --dataset mnist
python main.py --dataset cifar10

Results

MNIST

CIFAR-10

Reference

Sphere Generative Adversarial Network Based on Geometric Moment Matching

Sung Woo Park and Junseok Kwon. “Sphere Generative Adversarial Network Based on Geometric Moment Matching.” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2019.