Research and Analysis on Image Style Transfer Technologies

DOI

10.2991/978-94-6239-648-7_87How to use a DOI?

Keywords

Deep Learning; Generative Adversarial Networks; CycleGAN; Image Style Transfer

Abstract

Image style transfer technologies seek to imbue original images with a desired artistic style while preserving their content structure at the same time. Early image style transfer algorithms mostly employed non-parametric methods to achieve style transfer. Recently, there are a series of breakthroughs have been achieved relying on classic Generative Adversarial Networks (GANs). However, current mainstream methods remain to face multiple challenges in distinct areas such as general scene coverage, high-resolution control, domain-specific detail preservation, etc. This study centers on several classical GAN-based style translation strategies including DRB-GAN, DualStyleGAN, WeditGAN and Multi-scale CycleGAN. The central technique and algorithm system of image style transfer field mainly consists of these four mainstream technologies. In modern’s practices, these models show distinct limitations respectively in some aspects concerning the weak generalization of unknown styles, disappearance of non-facial details and troubles in real-time implementations. The future work conducted by scholars will seriously take challenges mentioned above into consideration and make progress in cross-modal fusion, detail augmentation designs and lightweight models. In summary, this review legibly demonstrates the different points on these models’ style preservation level, migration resilience and circumstances adaptability. This essay points out some advice on how to select reasonable technology in a particular scene and optimization directions for later researches in image style translation fields.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Runxin Yang
PY  - 2026
DA  - 2026/04/24
TI  - Research and Analysis on Image Style Transfer Technologies
BT  - Proceedings of the International Workshop on Advances in Deep Learning for Image Analysis and Computer Vision (IWADIC 2025)
PB  - Atlantis Press
SP  - 805
EP  - 814
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6239-648-7_87
DO  - 10.2991/978-94-6239-648-7_87
ID  - Yang2026
ER  -