Generative AI in Vision: A Survey on Models, Metrics and Applications

Gaurav Raut,Apoorv Singh

2024-02-26

Abstract:Generative AI models have revolutionized various fields by enabling the creation of realistic and diverse data samples. Among these models, diffusion models have emerged as a powerful approach for generating high-quality images, text, and audio. This survey paper provides a comprehensive overview of generative AI diffusion and legacy models, focusing on their underlying techniques, applications across different domains, and their challenges. We delve into the theoretical foundations of diffusion models, including concepts such as denoising diffusion probabilistic models (DDPM) and score-based generative modeling. Furthermore, we explore the diverse applications of these models in text-to-image, image inpainting, and image super-resolution, along with others, showcasing their potential in creative tasks and data augmentation. By synthesizing existing research and highlighting critical advancements in this field, this survey aims to provide researchers and practitioners with a comprehensive understanding of generative AI diffusion and legacy models and inspire future innovations in this exciting area of artificial intelligence.

Machine Learning,Artificial Intelligence

What problem does this paper attempt to address?

The problem this paper attempts to address is:

Generative AI models have made significant progress in the field of computer vision, especially in generating high-quality images, text, and audio. However, the theoretical foundations, application scope, and challenges faced by these models still require systematic summarization and analysis. This paper aims to provide a comprehensive overview of generative AI diffusion models and traditional models, focusing on their underlying technologies, cross-domain applications, and the challenges they face. Specifically, this paper delves into the theoretical foundations of diffusion models, including concepts such as Denoising Diffusion Probabilistic Models (DDPM) and score-based generative modeling, and explores the diverse applications of these models in tasks such as text-to-image generation, image restoration, and image super-resolution. By synthesizing existing research and highlighting key advancements, this paper aims to provide researchers and practitioners with a comprehensive understanding of generative AI diffusion models and traditional models, and to inspire future innovations in this field.

In short, the main objectives of this paper are:
1. To provide a comprehensive overview of the theoretical background and technical details of generative AI diffusion models.
2. To deeply analyze the applications and potential value of diffusion models in different fields.
3. To identify the shortcomings in existing research and propose future research directions to further advance the field.