ruifgf/monet-cyclegan

Some machine learning practice

(taken from Amy Jang's work at https://www.kaggle.com/code/amyjang/monet-cyclegan-tutorial)

CycleGAN

A Cycle-consistent Adversarial Networks attempts to map two domains by ensuring that the reciprocal mapping leads to the same result. This means that given a source and target domains (respectively, A and B) we are looking for mappings

such that

This model

The goal of this model is to take a photo (domain A) and "translate" it into an artistic style (domain B). We train this model with unpaired data, which means it is not comprised of "direct translations" between a given photo and the same with the expected artistic style.