Preprocessing

[openai]

Preprocessing

[voa18105]

I found the script 'generate tfr'. But it does not work from the box. What should be the structure of the dataset in order to generate tfrecords of it?

[prafullasd]

Which dataset are you trying to generate? Note you can directly downloading our processed tfr files using the links in the README, the scripts are only to see how we processed them.

lsun.py and imagenet_oord.py generate the tfr files for the datasets lsun, lsun_realnvp and imagenet-oord using the original images. Check the instructions at the top of the files on how to download the original images needed.

generate.py generates the tfr files for the datasets celeba-hq and imagenet using tfrecords files obtained from these scripts - https://github.com/tkarras/progressive_growing_of_gans/blob/master/dataset_tool.py
https://github.com/tensorflow/models/blob/master/research/inception/inception/data/build_imagenet_data.py