I am working on Pix2Pix GAN. Do we have a Pix2Pix GAN working facility on PerceptiLabs? If we have the facility of working with GAN please tell me how to do it?
It’s urgent !!!
Hi @VBB,
It looks like a Pix2Pix GAN is a cGAN (conditional GAN) which also contains a UNet as the generator.
The current version of PerceptiLabs unfortunately does not have any GANs available inside it, it’s missing the functionality to run two loss functions simultaneously as well as sharing layers between updates.
The older version of PL (0.11.15) does however have a GAN template available, but to make it a cGAN you would have to customize the training components code which is not the easiest thing to do (one of the reasons we replaced them).
So unless you want to try your hand on the older PL I would recommend to look elsewhere for a Pix2Pix GAN. This website seems to give a decent start: https://machinelearningmastery.com/how-to-develop-a-pix2pix-gan-for-image-to-image-translation/
Sorry that we can’t help more yet and hope your project goes well!