Research on Text Based Generative Image Editing

Yiren Wang

2024

Abstract

In recent years, with the continuous development of artificial intelligence the quality of images generated by using deep learning ability has gradually improved. Image editing uses the deep learning model to use conditional information as a guide and modify some regions in the image, while other regions remain unchanged. Generated image is a kind of automatic image editing, which can automatically generate images that meet the requirements of the user. Editable content includes, but is not limited to, the color, shape, texture, overall style and other features of the image to be edited. Generative image editing is an important part in the field of computer vision, and generative image editing has important use and theoretical value. This paper is a review based on the generative image editing with text as the control condition. This paper classifies the implementation method of text-based generative image editing, which is divided into generative adversarial network (GAN), diffusion model, CLIP model, and autoregressive model. The advantages and disadvantages of various implementation methods are analyzed, and various evaluation indexes of image editing are introduced. Finally, the future of the field is discussed.

Paper Citation

in Harvard Style

Wang Y. (2024). Research on Text Based Generative Image Editing. In Proceedings of the 1st International Conference on Modern Logistics and Supply Chain Management - Volume 1: MLSCM; ISBN 978-989-758-738-2, SciTePress, pages 146-150. DOI: 10.5220/0013241000004558

in Bibtex Style

@conference{mlscm24,
author={Yiren Wang},
title={Research on Text Based Generative Image Editing},
booktitle={Proceedings of the 1st International Conference on Modern Logistics and Supply Chain Management - Volume 1: MLSCM},
year={2024},
pages={146-150},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013241000004558},
isbn={978-989-758-738-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Modern Logistics and Supply Chain Management - Volume 1: MLSCM
TI - Research on Text Based Generative Image Editing
SN - 978-989-758-738-2
AU - Wang Y.
PY - 2024
SP - 146
EP - 150
DO - 10.5220/0013241000004558
PB - SciTePress