How is the dataset split into content and style set?

NVIDIA / FastPhotoStyle

Style transfer, deep learning, feature transform

Other

11.14k stars 1.2k forks source link

Closed ferric123 closed 5 years ago

ferric123 commented 5 years ago

Can I simply shuffle the dataset and in the training process randomly pick a pair of images as input?

mingyuliutw commented 5 years ago

Training of the model only happens at training the VGG decoders, which uses the COCO dataset.