A Review of Generative Adversarial Networks for Text to Image Tasks

Zihan Wo

2025

Abstract

To deal with the task of text-to-image generation, many models have been created in the past decade. In these models, Generative Adversarial Network (GAN) is a widely used basic model. Many models are developed based on GAN or some models are developed based on GAN. With the development of the research, the performance of models is getting better. From the vague and unreal images generated by the primitive models, to clear and reasonable images generated by newer models, the modeling of this task is gradually becoming more refined, and people’s understanding of this task is also being more completed. This paper will discuss the development process of the models by comparing several models with representative structures as a reference for subsequent researchers. Through this exploration, this paper aims to highlight the major developments and difficulties in text-to-image production, offering insights for future paths and possible enhancements in this quickly developing subject.

Paper Citation

in Harvard Style

Wo Z. (2025). A Review of Generative Adversarial Networks for Text to Image Tasks. In Proceedings of the 2nd International Conference on Data Science and Engineering - Volume 1: ICDSE; ISBN 978-989-758-765-8, SciTePress, pages 487-491. DOI: 10.5220/0013699800004670

in Bibtex Style

@conference{icdse25,
author={Zihan Wo},
title={A Review of Generative Adversarial Networks for Text to Image Tasks},
booktitle={Proceedings of the 2nd International Conference on Data Science and Engineering - Volume 1: ICDSE},
year={2025},
pages={487-491},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013699800004670},
isbn={978-989-758-765-8},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on Data Science and Engineering - Volume 1: ICDSE
TI - A Review of Generative Adversarial Networks for Text to Image Tasks
SN - 978-989-758-765-8
AU - Wo Z.
PY - 2025
SP - 487
EP - 491
DO - 10.5220/0013699800004670
PB - SciTePress