Generative Adversarial Network Technologies and Applications in Computer Vision.

Shengming Jiang,Fuxiao Tan,Lianchao Jin

doi:10.1155/2020/1459107

Shengming Jiang, Fuxiao Tan + Show 1 more

Open Access

https://doi.org/10.1155/2020/1459107

Copy DOI

Abstract

Computer vision is one of the hottest research fields in deep learning. The emergence of generative adversarial networks (GANs) provides a new method and model for computer vision. The idea of GANs using the game training method is superior to traditional machine learning algorithms in terms of feature learning and image generation. GANs are widely used not only in image generation and style transfer but also in the text, voice, video processing, and other fields. However, there are still some problems with GANs, such as model collapse and uncontrollable training. This paper deeply reviews the theoretical basis of GANs and surveys some recently developed GAN models, in comparison with traditional GAN models. The applications of GANs in computer vision include data enhancement, domain transfer, high-quality sample generation, and image restoration. The latest research progress of GANs in artificial intelligence (AI) based security attack and defense is introduced. The future development of GANs in computer vision is also discussed at the end of the paper with possible applications of AI in computer vision.

Highlights

Computer vision (CV) is a science that studies how to make a machine “see.” In 1963, Larry Roberts from MIT published the first doctoral dissertation in this field, “Machine Perception of ree-Dimensional Solids”, marking the beginning of CV research as a new direction of artificial intelligence. e background of this field was initially inspired by the human visual system, which is divided into two parts: the brain and eye. ey cooperate so that the human visual system can explain any scene
The real data is based on the biological signals: ECG and EEG, and the original dataset is extended by the generative adversarial model. e expansion of original data sets will make the classification model trained more effectively and improve the accuracy of classification. e expansion of data sets will help doctors improve the accuracy of diagnosis and determine the direction of treatment
We first review the latest research progress of computer vision, summarize the theoretical basis of generative adversarial networks (GANs) in detail, and elaborate on the challenges of GANs and the main advantages of GANs compared with traditional algorithms by combining its theoretical basis and practical use. e generative adversarial model has greatly promoted the rapid development of image processing field, with the continuous exploration of GAN models; they play an increasingly important role in other fields such as medicine, art, and security

Summary

Introduction

Computer vision (CV) is a science that studies how to make a machine “see.” In 1963, Larry Roberts from MIT published the first doctoral dissertation in this field, “Machine Perception of ree-Dimensional Solids”, marking the beginning of CV research as a new direction of artificial intelligence. e background of this field was initially inspired by the human visual system, which is divided into two parts: the brain and eye. ey cooperate so that the human visual system can explain any scene. With the rapid development of deep learning schemes, image restoration technology for missing information [7, 8] has achieved great success in recent years. Data enhancement technology [9, 10] has succeeded by generative models such as GANs. e increase of data volume makes the training effect of neural networks better and the generalization ability and robustness of models stronger, achieving higher accuracy in image recognition. Computer vision has been greatly promoted by the development of deep learning schemes, which have made great progress in the field of data processing [20, 21], promoting the rapid development and application of computer vision in many fields. The latest applications of GANs in computer vision are introduced, including data enhancement, high-quality sample generation, domain transfer, image restoration, and AI security. is paper is concluded in the fifth section by summarizing the development process of GANs and the future development direction of GANs in computer vision

Basic Theory of GANs

Evolution of GANs

Application of GANs in Computer Vision

Findings

Discussion and Conclusion