How to train a Generative model with a single image
Generative Adversorial Networks(GAN) are hot area of research in AI and they have a wide variety of applications like image to image, text to image translations, photos to emoji's, image editing etc. At the same time training process needs lot of computation power and huge amount of training data.
In my session i will be explaining about basic GANs and explain the architecture of SINGAN. and how we can train GAN using single image and will show how to use this trained GAN model applications like editing images, generate super resolution images, Creating animation from single image and i will be giving the code demo.
Outline/Structure of the Demonstration
- Introduction to GANs
- Applications of GANs
- Introduction to SINGAN
- How to train GAN with single image
- Applications of SINGAN
- Code demo
Attendees will learn about GANs and also will learn about new research paper in GAN i.e SINGAN
AI Researchers, Data Scientists, ML and DL enthusiasts
Prerequisites for Attendees
Basic knowledge about CNN's , Basic knowledge about GANs