TensorFlow Deep Learning Projects

Conditional GANs

In conditional GANs (CGANs), adding a vector of features controls the output and provides a better guide to the generator in figuring out what to do. Such a vector of features could encode the class the image should be derived be from (that is an image of a woman or a man if we are trying to create faces of imaginary actors) or even a set of specific characteristics we expect from the image (for imaginary actors, it could be the type of hair, eyes or complexion). The trick is done by incorporating the information into the images to be learned and into the Z input, which is not completely random anymore. The evaluation by the discriminator is done not only on the resemblance of fake data to the original data but also on the ...