Earticle

초록

영어

A lot of progress has happened since the seminal work of Gatys' neural style transfer. This survey aims to explain concepts of style transfer, and briefly look at various improvements after the IEEE survey published in 2020.

목차

Abstract
I. INTRODUCTION
II. BASIC EXPLANATION OF STYLE TRANSFER
A. Gatys' Neural Method
B. Loss Improvements
C. Approximating The Optimization Process by Neural Nets
D. WCT Variants
E. Why Match Statistics?
F. Patch Based Methods
G. Why VGG?
III. VARIOUS IMPROVEMENTS
A. Ablation Test
B. Content Leak Problem
C. Shape Modifying Style Transfer
D. Multi-Modal Style Transfer
E. Stroke Based Methods
F. Metric for Style Transfer
G. Style Transfer with Attention
H. Style Transfer with Image Translation
I. Other Improvements
IV. CONCLUSION
REFERENCES

저자

• Sungwoo Kang [ Electrical Engineering Korea University Seoul, Korea ] Corresponding Author
• Jonkook Kim [ Electrical Engineering Korea University Seoul, Korea ]

참고문헌