PEPSI++: Fast and Lightweight Network for Image Inpainting
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Shin, Y.-G. | - |
dc.contributor.author | Sagong, M.-C. | - |
dc.contributor.author | Yeo, Y.-J. | - |
dc.contributor.author | Kim, S.-W. | - |
dc.contributor.author | Ko, S.-J. | - |
dc.date.accessioned | 2021-08-30T02:19:04Z | - |
dc.date.available | 2021-08-30T02:19:04Z | - |
dc.date.created | 2021-06-17 | - |
dc.date.issued | 2021-01 | - |
dc.identifier.issn | 2162-237X | - |
dc.identifier.uri | https://scholar.korea.ac.kr/handle/2021.sw.korea/49393 | - |
dc.description.abstract | Among the various generative adversarial network (GAN)-based image inpainting methods, a coarse-to-fine network with a contextual attention module (CAM) has shown remarkable performance. However, due to two stacked generative networks, the coarse-to-fine network needs numerous computational resources, such as convolution operations and network parameters, which result in low speed. To address this problem, we propose a novel network architecture called parallel extended-decoder path for semantic inpainting (PEPSI) network, which aims at reducing the hardware costs and improving the inpainting performance. PEPSI consists of a single shared encoding network and parallel decoding networks called coarse and inpainting paths. The coarse path produces a preliminary inpainting result to train the encoding network for the prediction of features for the CAM. Simultaneously, the inpainting path generates higher inpainting quality using the refined features reconstructed via the CAM. In addition, we propose Diet-PEPSI that significantly reduces the network parameters while maintaining the performance. In Diet-PEPSI, to capture the global contextual information with low hardware costs, we propose novel rate-adaptive dilated convolutional layers that employ the common weights but produce dynamic features depending on the given dilation rates. Extensive experiments comparing the performance with state-of-the-art image inpainting methods demonstrate that both PEPSI and Diet-PEPSI improve the qualitative scores, i.e., the peak signal-to-noise ratio (PSNR) and structural similarity (SSIM), as well as significantly reduce hardware costs, such as computational time and the number of network parameters. © 2012 IEEE. | - |
dc.language | English | - |
dc.language.iso | en | - |
dc.publisher | Institute of Electrical and Electronics Engineers Inc. | - |
dc.subject | Convolution | - |
dc.subject | Cost reduction | - |
dc.subject | Decoding | - |
dc.subject | Encoding (symbols) | - |
dc.subject | Image enhancement | - |
dc.subject | Network architecture | - |
dc.subject | Semantics | - |
dc.subject | Signal to noise ratio | - |
dc.subject | Adversarial networks | - |
dc.subject | Computational resources | - |
dc.subject | Computational time | - |
dc.subject | Contextual information | - |
dc.subject | Low hardware costs | - |
dc.subject | Network parameters | - |
dc.subject | Peak signal to noise ratio | - |
dc.subject | Structural similarity | - |
dc.subject | Network coding | - |
dc.title | PEPSI++: Fast and Lightweight Network for Image Inpainting | - |
dc.type | Article | - |
dc.contributor.affiliatedAuthor | Ko, S.-J. | - |
dc.identifier.doi | 10.1109/TNNLS.2020.2978501 | - |
dc.identifier.scopusid | 2-s2.0-85099202814 | - |
dc.identifier.wosid | 000641162100021 | - |
dc.identifier.bibliographicCitation | IEEE Transactions on Neural Networks and Learning Systems, v.32, no.1, pp.252 - 265 | - |
dc.relation.isPartOf | IEEE Transactions on Neural Networks and Learning Systems | - |
dc.citation.title | IEEE Transactions on Neural Networks and Learning Systems | - |
dc.citation.volume | 32 | - |
dc.citation.number | 1 | - |
dc.citation.startPage | 252 | - |
dc.citation.endPage | 265 | - |
dc.type.rims | ART | - |
dc.type.docType | Article | - |
dc.description.journalClass | 1 | - |
dc.description.journalRegisteredClass | scie | - |
dc.description.journalRegisteredClass | scopus | - |
dc.relation.journalResearchArea | Computer Science | - |
dc.relation.journalResearchArea | Engineering | - |
dc.relation.journalWebOfScienceCategory | Computer Science, Artificial Intelligence | - |
dc.relation.journalWebOfScienceCategory | Computer Science, Hardware & Architecture | - |
dc.relation.journalWebOfScienceCategory | Computer Science, Theory & Methods | - |
dc.relation.journalWebOfScienceCategory | Engineering, Electrical & Electronic | - |
dc.subject.keywordPlus | Convolution | - |
dc.subject.keywordPlus | Cost reduction | - |
dc.subject.keywordPlus | Decoding | - |
dc.subject.keywordPlus | Encoding (symbols) | - |
dc.subject.keywordPlus | Image enhancement | - |
dc.subject.keywordPlus | Network architecture | - |
dc.subject.keywordPlus | Semantics | - |
dc.subject.keywordPlus | Signal to noise ratio | - |
dc.subject.keywordPlus | Adversarial networks | - |
dc.subject.keywordPlus | Computational resources | - |
dc.subject.keywordPlus | Computational time | - |
dc.subject.keywordPlus | Contextual information | - |
dc.subject.keywordPlus | Low hardware costs | - |
dc.subject.keywordPlus | Network parameters | - |
dc.subject.keywordPlus | Peak signal to noise ratio | - |
dc.subject.keywordPlus | Structural similarity | - |
dc.subject.keywordPlus | Network coding | - |
dc.subject.keywordAuthor | Deep learning | - |
dc.subject.keywordAuthor | generative adversarial network (GAN) | - |
dc.subject.keywordAuthor | image inpainting | - |
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.
(02841) 서울특별시 성북구 안암로 14502-3290-1114
COPYRIGHT © 2021 Korea University. All Rights Reserved.
Certain data included herein are derived from the © Web of Science of Clarivate Analytics. All rights reserved.
You may not copy or re-distribute this material in whole or in part without the prior written consent of Clarivate Analytics.