no code implementations • 19 Dec 2018 • Holly Grimm
I consider how to influence CycleGAN, image-to-image translation, by using additional constraints from a neural network trained on art composition attributes.
Attribute Image-to-Image Translation +1