BigGAN Generators with Pretrained Weights in Pytorch

Pytorch implementation of the generator of Large Scale GAN Training for High Fidelity Natural Image Synthesis (BigGAN).

Download Pretrained Weights

The Pretrained weights can be downloaded from the latest release. link

Dependencies

dependencies:

• python=3.6
• cudatoolkit=10.0
• pytorch
• torchvision
• scipy

Please also refer to the environment.yml file.

Demo

• To run the code, please download the pretrained weights first.

python demo.py -w <PRETRAINED_WEIGHT_PATH> [-s IMAGE_SIZE] [-c CLASS_LABEL] [-t TRUNCATION] 
python demo.py -w ./biggan512-release.pt -s 512 -t 0.3 -c 156 
python demo.py -w ./biggan256-release.pt -s 256 -t 0.02 -c 11 
python demo.py --pretrained_weight ./biggan128-release.pt --size 128 --truncation 0.2 --class_label 821 

• Valid image size: 128, 256, 512
• Valid class label: 0~999
• Valid truncation: 0.02~1.0

Results

class 156 (512 x 512)
class 11 (512 x 512)
class 821 (512 x 512)

Pretrained Weights

The pretrained weights are converted from the tensorflow hub modules:

• https://tfhub.dev/deepmind/biggan-128/2
• https://tfhub.dev/deepmind/biggan-256/2
• https://tfhub.dev/deepmind/biggan-512/2

References

paper: https://arxiv.org/abs/1809.11096

https://github.com/ajbrock/BigGAN-PyTorch

Contact

Please feel free to leave suggestions or comments to Tsung-Hung Hsieh(andrewhsiehth@gmail.com).