Deep learning vs. traditional algorithms for saliency prediction of distorted images

Xin Zhao; Hanhe Lin; Pengfei Guo; Dietmar Saupe; Hantao Liu

doi:10.1109/ICIP40778.2020.9191203

Deep learning vs. traditional algorithms for saliency prediction of distorted images

Xin Zhao, Hanhe Lin, Pengfei Guo, Dietmar Saupe, Hantao Liu

Computing

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

4 Citations (Scopus)

Abstract

Saliency has been widely studied in relation to image quality assessment (IQA). The optimal use of saliency in IQA metrics, however, is nontrivial and largely depends on whether saliency can be accurately predicted for images containing various distortions. Although tremendous progress has been made in saliency modelling, very little is known about whether and to what extent state-of-the-art methods are beneficial for saliency prediction of distorted images. In this paper, we analyse the ability of deep learning versus traditional algorithms in predicting saliency, based on an IQA-aware saliency benchmark, the SIQ288 database. Building off the variations in model performance, we make recommendations for model selections for IQA applications.

Original language	English
Title of host publication	2020 IEEE International Conference on Image Processing (ICIP)
Subtitle of host publication	proceedings
Publisher	IEEE
Pages	156-160
Number of pages	5
ISBN (Electronic)	978-1-7281-6395-6
ISBN (Print)	978-1-7281-6396-3
DOIs	https://doi.org/10.1109/ICIP40778.2020.9191203
Publication status	Published - Oct 2020
Event	2020 IEEE International Conference on Image Processing - Virtual, Abu Dhabi, United Arab Emirates Duration: 25 Sept 2020 → 28 Sept 2020

Publication series

Name	Proceedings - International Conference on Image Processing, ICIP
Volume	2020-October
ISSN (Print)	1522-4880

Conference

Conference	2020 IEEE International Conference on Image Processing
Abbreviated title	ICIP
Country/Territory	United Arab Emirates
City	Virtual, Abu Dhabi
Period	25/09/20 → 28/09/20

Keywords

distortion
eyetracking
Image quality assessment
saliency
statistical analysis

ASJC Scopus subject areas

Software
Computer Vision and Pattern Recognition
Signal Processing

Access to Document

10.1109/ICIP40778.2020.9191203

Cite this

Zhao, X., Lin, H., Guo, P., Saupe, D., & Liu, H. (2020). Deep learning vs. traditional algorithms for saliency prediction of distorted images. In 2020 IEEE International Conference on Image Processing (ICIP): proceedings (pp. 156-160). (Proceedings - International Conference on Image Processing, ICIP; Vol. 2020-October). IEEE. https://doi.org/10.1109/ICIP40778.2020.9191203

@inproceedings{e3211ab7bbac4d3ba825fd9c1d37c2a2,

title = "Deep learning vs. traditional algorithms for saliency prediction of distorted images",

abstract = "Saliency has been widely studied in relation to image quality assessment (IQA). The optimal use of saliency in IQA metrics, however, is nontrivial and largely depends on whether saliency can be accurately predicted for images containing various distortions. Although tremendous progress has been made in saliency modelling, very little is known about whether and to what extent state-of-the-art methods are beneficial for saliency prediction of distorted images. In this paper, we analyse the ability of deep learning versus traditional algorithms in predicting saliency, based on an IQA-aware saliency benchmark, the SIQ288 database. Building off the variations in model performance, we make recommendations for model selections for IQA applications.",

keywords = "distortion, eyetracking, Image quality assessment, saliency, statistical analysis",

author = "Xin Zhao and Hanhe Lin and Pengfei Guo and Dietmar Saupe and Hantao Liu",

note = "Funding Information: Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project-ID 251654672 – TRR 161 (Project A05). Publisher Copyright: {\textcopyright} 2020 IEEE.; 2020 IEEE International Conference on Image Processing<br/>, ICIP ; Conference date: 25-09-2020 Through 28-09-2020",

year = "2020",

month = oct,

doi = "10.1109/ICIP40778.2020.9191203",

language = "English",

isbn = "978-1-7281-6396-3",

series = "Proceedings - International Conference on Image Processing, ICIP",

publisher = "IEEE",

pages = "156--160",

booktitle = "2020 IEEE International Conference on Image Processing (ICIP)",

}

Zhao, X, Lin, H, Guo, P, Saupe, D & Liu, H 2020, Deep learning vs. traditional algorithms for saliency prediction of distorted images. in 2020 IEEE International Conference on Image Processing (ICIP): proceedings. Proceedings - International Conference on Image Processing, ICIP, vol. 2020-October, IEEE, pp. 156-160, 2020 IEEE International Conference on Image Processing
, Virtual, Abu Dhabi, United Arab Emirates, 25/09/20. https://doi.org/10.1109/ICIP40778.2020.9191203

Deep learning vs. traditional algorithms for saliency prediction of distorted images. / Zhao, Xin; Lin, Hanhe; Guo, Pengfei et al.
2020 IEEE International Conference on Image Processing (ICIP): proceedings. IEEE, 2020. p. 156-160 (Proceedings - International Conference on Image Processing, ICIP; Vol. 2020-October).

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

TY - GEN

T1 - Deep learning vs. traditional algorithms for saliency prediction of distorted images

AU - Zhao, Xin

AU - Lin, Hanhe

AU - Guo, Pengfei

AU - Saupe, Dietmar

AU - Liu, Hantao

PY - 2020/10

Y1 - 2020/10

N2 - Saliency has been widely studied in relation to image quality assessment (IQA). The optimal use of saliency in IQA metrics, however, is nontrivial and largely depends on whether saliency can be accurately predicted for images containing various distortions. Although tremendous progress has been made in saliency modelling, very little is known about whether and to what extent state-of-the-art methods are beneficial for saliency prediction of distorted images. In this paper, we analyse the ability of deep learning versus traditional algorithms in predicting saliency, based on an IQA-aware saliency benchmark, the SIQ288 database. Building off the variations in model performance, we make recommendations for model selections for IQA applications.

AB - Saliency has been widely studied in relation to image quality assessment (IQA). The optimal use of saliency in IQA metrics, however, is nontrivial and largely depends on whether saliency can be accurately predicted for images containing various distortions. Although tremendous progress has been made in saliency modelling, very little is known about whether and to what extent state-of-the-art methods are beneficial for saliency prediction of distorted images. In this paper, we analyse the ability of deep learning versus traditional algorithms in predicting saliency, based on an IQA-aware saliency benchmark, the SIQ288 database. Building off the variations in model performance, we make recommendations for model selections for IQA applications.

KW - distortion

KW - eyetracking

KW - Image quality assessment

KW - saliency

KW - statistical analysis

UR - http://www.scopus.com/inward/record.url?scp=85098625821&partnerID=8YFLogxK

U2 - 10.1109/ICIP40778.2020.9191203

DO - 10.1109/ICIP40778.2020.9191203

M3 - Conference contribution

AN - SCOPUS:85098625821

SN - 978-1-7281-6396-3

T3 - Proceedings - International Conference on Image Processing, ICIP

SP - 156

EP - 160

BT - 2020 IEEE International Conference on Image Processing (ICIP)

PB - IEEE

T2 - 2020 IEEE International Conference on Image Processing<br/>

Y2 - 25 September 2020 through 28 September 2020

ER -

Deep learning vs. traditional algorithms for saliency prediction of distorted images

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this