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

Manri Cheon; Jun Hyuk Kim; Jun Ho Choi; Jong Seok Lee

doi:10.1007/978-3-030-11021-5_4

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

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

School of Integrated Technology

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

9 Citations (Scopus)

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 language	English
Title of host publication	Computer Vision – ECCV 2018 Workshops, Proceedings
Editors	Laura Leal-Taixé, Stefan Roth
Publisher	Springer Verlag
Pages	51-62
Number of pages	12
ISBN (Print)	9783030110208
DOIs	https://doi.org/10.1007/978-3-030-11021-5_4
Publication status	Published - 2019
Event	15th European Conference on Computer Vision, ECCV 2018 - Munich, Germany Duration: 2018 Sept 8 → 2018 Sept 14

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11133 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Other

Other	15th European Conference on Computer Vision, ECCV 2018
Country/Territory	Germany
City	Munich
Period	18/9/8 → 18/9/14

Bibliographical note

Publisher Copyright:
© Springer Nature Switzerland AG 2019.

All Science Journal Classification (ASJC) codes

Theoretical Computer Science
General Computer Science

Access to Document

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

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 et al. / 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.",

keywords = "Deep learning, Distortion, Perception, Super-resolution",

author = "Manri Cheon and Kim, {Jun Hyuk} and Choi, {Jun Ho} and Lee, {Jong Seok}",

note = "Publisher Copyright: {\textcopyright} Springer Nature Switzerland AG 2019.; 15th European Conference on Computer Vision, ECCV 2018 ; Conference date: 08-09-2018 Through 14-09-2018",

year = "2019",

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, JS 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 et al.
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 proceeding › Conference 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

N1 - Publisher Copyright: © Springer Nature Switzerland AG 2019.

PY - 2019

Y1 - 2019

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.

KW - Deep learning

KW - Distortion

KW - Perception

KW - Super-resolution

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

T2 - 15th European Conference on Computer Vision, ECCV 2018

Y2 - 8 September 2018 through 14 September 2018

ER -

Cheon M, Kim JH, Choi JH, Lee JS. 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)). doi: 10.1007/978-3-030-11021-5_4

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

Abstract

Publication series

Other

Bibliographical note

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this