Say I am using a CycleGAN to generate Zebra from Horse. But for my case, there is also deer present in both of the images, and I don't want to change the deer from my input and keep them the same in output, only changing horse to zebra.

So, I guess I have to mask the deer in my training label so that the generator does not learn to change it. How can I achieve that? Does anyone have any idea? (Let's say I have deer/not-deer mask for all my training data)



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.