Research and application of image inpainting based on vision transformer and generative adversarial networks

Xingyu Lu; Luqiao Zhang; Yi Xiang; Shasha Mo

doi:10.1117/12.3004149

11 October 2023 Research and application of image inpainting based on vision transformer and generative adversarial networks

Xingyu Lu, Luqiao Zhang, Yi Xiang, Shasha Mo

Author Affiliations +

Proceedings Volume 12800, Sixth International Conference on Computer Information Science and Application Technology (CISAT 2023); 128004V (2023) https://doi.org/10.1117/12.3004149
Event: 6th International Conference on Computer Information Science and Application Technology (CISAT 2023), 2023, Hangzhou, China

Abstract

Image inpainting techniques have gradually been replaced by deep learning-based methods from traditional methods based on structure and texture, and the effectiveness has been improved. However, there are still many issues, such as the problem of blurring, color distortion, and inconsistent semantic information with the intact region in the restored image. This paper proposes a method that combines Vision Transformer and Generative Adversarial Networks, and trains the model with large-scale masks to enable the Transformer encoder to learn deep-level feature information as much as possible. Traditional convolution cannot make the model dynamically distinguish valid information from invalid information. In this method, gated convolution is added to the generator to mask out invalid information in missing areas. Compared with the traditional attention model, the method proposed in this paper has better structural texture and semantic consistency in the inpainting effect.

(2023) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Xingyu Lu, Luqiao Zhang, Yi Xiang, and Shasha Mo "Research and application of image inpainting based on vision transformer and generative adversarial networks", Proc. SPIE 12800, Sixth International Conference on Computer Information Science and Application Technology (CISAT 2023), 128004V (11 October 2023); https://doi.org/10.1117/12.3004149

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

;

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available

Members: $17.00

Non-members: $21.00 ADD TO CART

PROCEEDINGS
5 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Image restoration

Visual process modeling

Convolution

Convolutional neural networks

Transformers

Education and training

Semantics

Show All Keywords

Keywords/Phrases

Search In:

Publication Years