Free-Form Mask in Image Inpainting with GC-PatchGAN for High-Resolution Natural Scenery

Xuanze Chen

2023

Abstract

The research delves into the domain of image inpainting, an essential task in image processing with widespread applications. Image inpainting holds immense importance in restoring damaged or incomplete images, finding utility in photography, video processing, and other fields. The study aims to introduce an innovative approach that combines the Contextual Attention Layer, Gated Convolution and SN-PatchGAN (GC-PatchGAN) to enhance inpainting outcomes. The methodology incorporates the Contextual Attention Layer for strategic borrowing of feature information from known background patches. Gated Convolution dynamically selects and highlights pertinent features, leading to a substantial improvement in inpainting quality. Extensive experiments were conducted to assess the proposed method, utilizing the Kaggle dataset. The results consistently demonstrate exceptional performance across various scenarios, including those involving Free-Form masks and user-guided input. Gated Convolution plays a pivotal role in generating high-quality results consistently. In practical terms, this research contributes significantly to image restoration, facilitating the removal of distractions, layout modifications, watermark elimination, and facial editing within images. Additionally, it addresses the challenge of irregular objects obstructing landscapes in photographs. In conclusion, this study advances the field of image inpainting, holding considerable promise for enhancing image quality and editing capabilities across diverse industries reliant on image processing and restoration.

Paper Citation

in Harvard Style

Chen X. (2023). Free-Form Mask in Image Inpainting with GC-PatchGAN for High-Resolution Natural Scenery. In Proceedings of the 1st International Conference on Data Analysis and Machine Learning - Volume 1: DAML; ISBN 978-989-758-705-4, SciTePress, pages 402-406. DOI: 10.5220/0012819300003885

in Bibtex Style

@conference{daml23,
author={Xuanze Chen},
title={Free-Form Mask in Image Inpainting with GC-PatchGAN for High-Resolution Natural Scenery},
booktitle={Proceedings of the 1st International Conference on Data Analysis and Machine Learning - Volume 1: DAML},
year={2023},
pages={402-406},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012819300003885},
isbn={978-989-758-705-4},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Data Analysis and Machine Learning - Volume 1: DAML
TI - Free-Form Mask in Image Inpainting with GC-PatchGAN for High-Resolution Natural Scenery
SN - 978-989-758-705-4
AU - Chen X.
PY - 2023
SP - 402
EP - 406
DO - 10.5220/0012819300003885
PB - SciTePress