Training on Art Composition Attributes to Influence CycleGAN Art Generation
This work addresses the challenge of controlling artistic style in generative models for image translation, but it appears incremental as it builds on existing CycleGAN methods with added constraints.
The authors tackled the problem of influencing CycleGAN-based image-to-image translation by incorporating art composition attributes, training a neural network (ACAN) using domain knowledge from art evaluation rules and applying it to apple2orange translation.
I consider how to influence CycleGAN, image-to-image translation, by using additional constraints from a neural network trained on art composition attributes. I show how I trained the the Art Composition Attributes Network (ACAN) by incorporating domain knowledge based on the rules of art evaluation and the result of applying each art composition attribute to apple2orange image translation.