Related references
Note: Only part of the references are listed.
Proceedings Paper
Computer Science, Artificial Intelligence
Roman Suvorov et al.
Summary: Modern image inpainting systems often struggle with large missing areas, complex geometric structures, and high-resolution images. This study proposes a new method called large mask inpainting (LaMa), which improves the inpainting network architecture using fast Fourier convolutions (FFCs) with wide receptive fields and a high receptive field perceptual loss. The experiments show that the proposed method achieves excellent performance and can handle challenging scenarios effectively.
2022 IEEE WINTER CONFERENCE ON APPLICATIONS OF COMPUTER VISION (WACV 2022)
(2022)
Article
Computer Science, Artificial Intelligence
Weize Quan et al.
Summary: This study rethinks the process of image inpainting from the perspective of receptive field and proposes a three-stage inpainting framework with local and global refinement. Experimental results demonstrate that the proposed method outperforms existing techniques on image inpainting tasks.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Article
Engineering, Electrical & Electronic
Bin Li et al.
Summary: This research proposes a deep learning approach for image inpainting, which fills semantic content and enhances details in both spatial and wavelet domains to improve restoration quality. Experimental results demonstrate the superiority of this method over current state-of-the-art ones.
Article
Engineering, Electrical & Electronic
Hiya Roy et al.
Summary: This paper introduces an image inpainting technique using frequency-domain information, which outperforms current state-of-the-art methods by training networks to better reconstruct high-frequency details.
JOURNAL OF ELECTRONIC IMAGING
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Yingchen Yu et al.
Summary: The paper introduces a wavelet-based inpainting method that decomposes images into multiple frequency bands and fills missing regions with different loss functions for each band, effectively mitigating interfrequency conflicts and achieving superior image inpainting results.
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021)
(2021)
Article
Computer Science, Artificial Intelligence
Ning Wang et al.
Summary: This paper introduces a novel dynamic selection network (DSNet) to address the challenge of distinguishing between corrupted regions and valid regions in image inpainting tasks, aiming to improve the accuracy and realism of image inpainting. DSNet utilizes the validness migratable convolution (VMC) and regional composite normalization (RCN) modules, employing dynamic selection mechanisms and adaptive feature extraction processes to achieve better image inpainting results.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2021)
Article
Computer Science, Information Systems
Xingchen He et al.
Proceedings Paper
Computer Science, Artificial Intelligence
Phillip Isola et al.
30TH IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2017)
(2017)
Proceedings Paper
Computer Science, Artificial Intelligence
Mircea Cimpoi et al.
2014 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)
(2014)
Article
Computer Science, Software Engineering
Connelly Barnes et al.
ACM TRANSACTIONS ON GRAPHICS
(2009)
Article
Computer Science, Artificial Intelligence
Z Wang et al.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2004)
Article
Computer Science, Artificial Intelligence
A Criminisi et al.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2004)
Article
Mathematics, Applied
TF Chan et al.
SIAM JOURNAL ON APPLIED MATHEMATICS
(2002)
