Generative AI From Theory to Model

Abstract

This book chapter provides a comprehensive overview of generative AI and its applications in computer vision. The introduction section elucidates the concept of generative AI and underscores its importance within the realm of artificial intelligence. The chapter also provides a deep dive into the various techniques used in generative AI, such as creative style transfer, forecasting subsequent video frames, enhancing image resolution, enabling interactive image generation, facilitating image-to-image translation, text-to-image synthesis, image inpainting, the generation of innovative animated characters, the construction of 3D models from image, the utilization of the variational autoencoder (VAE) and its various adaptations, the implementation of generative adversarial networks (GANs) and their diverse iterations, as well as the use of transformers and their manifold versions. The chapter also highlights the current limitations and potential future developments in the field.