Generative adversarial network-based image super-resolution using perceptual content losses

Manri Cheon, Jun Hyuk Kim, Jun Ho Choi, Jong-Seok Lee

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In this paper, we propose a deep generative adversarial network for super-resolution considering the trade-off between perception and distortion. Based on good performance of a recently developed model for super-resolution, i.e., deep residual network using enhanced upscale modules (EUSR) [9], the proposed model is trained to improve perceptual performance with only slight increase of distortion. For this purpose, together with the conventional content loss, i.e., reconstruction loss such as L1 or L2, we consider additional losses in the training phase, which are the discrete cosine transform coefficients loss and differential content loss. These consider perceptual part in the content loss, i.e., consideration of proper high frequency components is helpful for the trade-off problem in super-resolution. The experimental results show that our proposed model has good performance for both perception and distortion, and is effective in perceptual super-resolution applications.

Original languageEnglish
Title of host publicationComputer Vision – ECCV 2018 Workshops, Proceedings
EditorsLaura Leal-Taixé, Stefan Roth
PublisherSpringer Verlag
Pages51-62
Number of pages12
ISBN (Print)9783030110208
DOIs
Publication statusPublished - 2019 Jan 1
Event15th European Conference on Computer Vision, ECCV 2018 - Munich, Germany
Duration: 2018 Sep 82018 Sep 14

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11133 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other15th European Conference on Computer Vision, ECCV 2018
CountryGermany
CityMunich
Period18/9/818/9/14

Fingerprint

Super-resolution
Trade-offs
Discrete Cosine Transform
Discrete cosine transforms
Model
Module
Experimental Results
Coefficient

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Cheon, M., Kim, J. H., Choi, J. H., & Lee, J-S. (2019). Generative adversarial network-based image super-resolution using perceptual content losses. In L. Leal-Taixé, & S. Roth (Eds.), Computer Vision – ECCV 2018 Workshops, Proceedings (pp. 51-62). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11133 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-11021-5_4
Cheon, Manri ; Kim, Jun Hyuk ; Choi, Jun Ho ; Lee, Jong-Seok. / Generative adversarial network-based image super-resolution using perceptual content losses. Computer Vision – ECCV 2018 Workshops, Proceedings. editor / Laura Leal-Taixé ; Stefan Roth. Springer Verlag, 2019. pp. 51-62 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{13566daa3a2d44b0827bd8203f40f5b5,
title = "Generative adversarial network-based image super-resolution using perceptual content losses",
abstract = "In this paper, we propose a deep generative adversarial network for super-resolution considering the trade-off between perception and distortion. Based on good performance of a recently developed model for super-resolution, i.e., deep residual network using enhanced upscale modules (EUSR) [9], the proposed model is trained to improve perceptual performance with only slight increase of distortion. For this purpose, together with the conventional content loss, i.e., reconstruction loss such as L1 or L2, we consider additional losses in the training phase, which are the discrete cosine transform coefficients loss and differential content loss. These consider perceptual part in the content loss, i.e., consideration of proper high frequency components is helpful for the trade-off problem in super-resolution. The experimental results show that our proposed model has good performance for both perception and distortion, and is effective in perceptual super-resolution applications.",
author = "Manri Cheon and Kim, {Jun Hyuk} and Choi, {Jun Ho} and Jong-Seok Lee",
year = "2019",
month = "1",
day = "1",
doi = "10.1007/978-3-030-11021-5_4",
language = "English",
isbn = "9783030110208",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "51--62",
editor = "Laura Leal-Taix{\'e} and Stefan Roth",
booktitle = "Computer Vision – ECCV 2018 Workshops, Proceedings",
address = "Germany",

}

Cheon, M, Kim, JH, Choi, JH & Lee, J-S 2019, Generative adversarial network-based image super-resolution using perceptual content losses. in L Leal-Taixé & S Roth (eds), Computer Vision – ECCV 2018 Workshops, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11133 LNCS, Springer Verlag, pp. 51-62, 15th European Conference on Computer Vision, ECCV 2018, Munich, Germany, 18/9/8. https://doi.org/10.1007/978-3-030-11021-5_4

Generative adversarial network-based image super-resolution using perceptual content losses. / Cheon, Manri; Kim, Jun Hyuk; Choi, Jun Ho; Lee, Jong-Seok.

Computer Vision – ECCV 2018 Workshops, Proceedings. ed. / Laura Leal-Taixé; Stefan Roth. Springer Verlag, 2019. p. 51-62 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11133 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Generative adversarial network-based image super-resolution using perceptual content losses

AU - Cheon, Manri

AU - Kim, Jun Hyuk

AU - Choi, Jun Ho

AU - Lee, Jong-Seok

PY - 2019/1/1

Y1 - 2019/1/1

N2 - In this paper, we propose a deep generative adversarial network for super-resolution considering the trade-off between perception and distortion. Based on good performance of a recently developed model for super-resolution, i.e., deep residual network using enhanced upscale modules (EUSR) [9], the proposed model is trained to improve perceptual performance with only slight increase of distortion. For this purpose, together with the conventional content loss, i.e., reconstruction loss such as L1 or L2, we consider additional losses in the training phase, which are the discrete cosine transform coefficients loss and differential content loss. These consider perceptual part in the content loss, i.e., consideration of proper high frequency components is helpful for the trade-off problem in super-resolution. The experimental results show that our proposed model has good performance for both perception and distortion, and is effective in perceptual super-resolution applications.

AB - In this paper, we propose a deep generative adversarial network for super-resolution considering the trade-off between perception and distortion. Based on good performance of a recently developed model for super-resolution, i.e., deep residual network using enhanced upscale modules (EUSR) [9], the proposed model is trained to improve perceptual performance with only slight increase of distortion. For this purpose, together with the conventional content loss, i.e., reconstruction loss such as L1 or L2, we consider additional losses in the training phase, which are the discrete cosine transform coefficients loss and differential content loss. These consider perceptual part in the content loss, i.e., consideration of proper high frequency components is helpful for the trade-off problem in super-resolution. The experimental results show that our proposed model has good performance for both perception and distortion, and is effective in perceptual super-resolution applications.

UR - http://www.scopus.com/inward/record.url?scp=85061709563&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85061709563&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-11021-5_4

DO - 10.1007/978-3-030-11021-5_4

M3 - Conference contribution

AN - SCOPUS:85061709563

SN - 9783030110208

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 51

EP - 62

BT - Computer Vision – ECCV 2018 Workshops, Proceedings

A2 - Leal-Taixé, Laura

A2 - Roth, Stefan

PB - Springer Verlag

ER -

Cheon M, Kim JH, Choi JH, Lee J-S. Generative adversarial network-based image super-resolution using perceptual content losses. In Leal-Taixé L, Roth S, editors, Computer Vision – ECCV 2018 Workshops, Proceedings. Springer Verlag. 2019. p. 51-62. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-11021-5_4