GANs for Data Augmentation and Privacy

Learning Objectives

• Explore different applications of GANs and examine their specific applications in data augmentation, privacy, and anonymity.
• Improve your downstream AI models with GAN-generated data.

Image To Image Translation with Pix2Pix

Learning Objectives

• Learn and leverage the image-to-image translation framework.
• Identify extensions, generalizations, and applications of this framework to modalities beyond images.
• Implement a paired image-to-image translation GAN, called Pix2Pix, to turn satellite images into map routes (and vice versa), with advanced U-Net generator and PatchGAN discriminator architectures.

Unpaired Translation with CycleGAN

Learning Objectives

• Compare paired image-to-image translation to unpaired image-to-image translation.
• Identify how their key difference necessitates a different GAN architecture.
• Implement unpaired image-to-image translation model, called CycleGAN, to adapt horses to zebras (and vice versa) with two GANs in one.