Discriminator Synthesis: On reusing the other half of Generative Adversarial Networks
This work addresses a specific inefficiency in GANs for computer vision researchers, offering a novel application of existing components, though it is incremental in scope.
The paper tackles the underutilization of the Discriminator in Generative Adversarial Networks after training by proposing Discriminator Dreaming, a method that uses its learned features to alter and generate images from scratch, achieving competitive results in image manipulation tasks.
Generative Adversarial Networks have long since revolutionized the world of computer vision and, tied to it, the world of art. Arduous efforts have gone into fully utilizing and stabilizing training so that outputs of the Generator network have the highest possible fidelity, but little has gone into using the Discriminator after training is complete. In this work, we propose to use the latter and show a way to use the features it has learned from the training dataset to both alter an image and generate one from scratch. We name this method Discriminator Dreaming, and the full code can be found at https://github.com/PDillis/stylegan3-fun.