Co-saliency detection based on siamese network

Zhengchao Lei; Weiyan Chai; Sanyuan Zhao; Hongmei Song; Fengxia Li

doi:10.1007/978-981-10-8890-2_8

Co-saliency detection based on siamese network

Zhengchao Lei, Weiyan Chai, Sanyuan Zhao^*, Hongmei Song, Fengxia Li

^*Corresponding author for this work

School of Computer Science and Technology

Beijing Institute of Technology

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

Abstract

Saliency detection in images attracts much research attention for its usage in numerous multimedia applications. Beside on the detection within the single image, co-saliency has been developed rapidly by detecting the same foreground objects in different images and trying to further promote the performance of object detection. This paper we propose a co-saliency detection method based on Siamese Network. By using Siamese Network, we get the similarity matrix of each image in superpixels. Guided by the single image saliency map, each saliency value, saliency score matrix is obtained to generate the multi image saliency map. Our final saliency map is a linear combination of these two saliency maps. The experiments show that our method performs better than other state-of-arts methods.

Original language	English
Title of host publication	Mobile Ad-hoc and Sensor Networks - 13th International Conference, MSN 2017, Revised Selected Papers
Editors	Liehuang Zhu, Sheng Zhong
Publisher	Springer Verlag
Pages	99-109
Number of pages	11
ISBN (Print)	9789811088896
DOIs	https://doi.org/10.1007/978-981-10-8890-2_8
Publication status	Published - 2018
Event	13th International Conference on Mobile Ad-hoc and Sensor Networks, MSN 2017 - Beijing, China Duration: 17 Dec 2017 → 20 Dec 2017

Publication series

Name	Communications in Computer and Information Science
Volume	747
ISSN (Print)	1865-0929

Conference

Conference	13th International Conference on Mobile Ad-hoc and Sensor Networks, MSN 2017
Country/Territory	China
City	Beijing
Period	17/12/17 → 20/12/17

Keywords

Co-saliency detection
Feature extraction
Siamese network

Access to Document

10.1007/978-981-10-8890-2_8

Cite this

Lei, Z., Chai, W., Zhao, S., Song, H., & Li, F. (2018). Co-saliency detection based on siamese network. In L. Zhu, & S. Zhong (Eds.), Mobile Ad-hoc and Sensor Networks - 13th International Conference, MSN 2017, Revised Selected Papers (pp. 99-109). (Communications in Computer and Information Science; Vol. 747). Springer Verlag. https://doi.org/10.1007/978-981-10-8890-2_8

@inproceedings{119665fed66b4796b427e3734ca01064,

title = "Co-saliency detection based on siamese network",

abstract = "Saliency detection in images attracts much research attention for its usage in numerous multimedia applications. Beside on the detection within the single image, co-saliency has been developed rapidly by detecting the same foreground objects in different images and trying to further promote the performance of object detection. This paper we propose a co-saliency detection method based on Siamese Network. By using Siamese Network, we get the similarity matrix of each image in superpixels. Guided by the single image saliency map, each saliency value, saliency score matrix is obtained to generate the multi image saliency map. Our final saliency map is a linear combination of these two saliency maps. The experiments show that our method performs better than other state-of-arts methods.",

keywords = "Co-saliency detection, Feature extraction, Siamese network",

author = "Zhengchao Lei and Weiyan Chai and Sanyuan Zhao and Hongmei Song and Fengxia Li",

note = "Publisher Copyright: {\textcopyright} Springer Nature Singapore Pte Ltd. 2018.; 13th International Conference on Mobile Ad-hoc and Sensor Networks, MSN 2017 ; Conference date: 17-12-2017 Through 20-12-2017",

year = "2018",

doi = "10.1007/978-981-10-8890-2_8",

language = "English",

isbn = "9789811088896",

series = "Communications in Computer and Information Science",

publisher = "Springer Verlag",

pages = "99--109",

editor = "Liehuang Zhu and Sheng Zhong",

booktitle = "Mobile Ad-hoc and Sensor Networks - 13th International Conference, MSN 2017, Revised Selected Papers",

address = "Germany",

}

Lei, Z, Chai, W, Zhao, S, Song, H & Li, F 2018, Co-saliency detection based on siamese network. in L Zhu & S Zhong (eds), Mobile Ad-hoc and Sensor Networks - 13th International Conference, MSN 2017, Revised Selected Papers. Communications in Computer and Information Science, vol. 747, Springer Verlag, pp. 99-109, 13th International Conference on Mobile Ad-hoc and Sensor Networks, MSN 2017, Beijing, China, 17/12/17. https://doi.org/10.1007/978-981-10-8890-2_8

Co-saliency detection based on siamese network. / Lei, Zhengchao; Chai, Weiyan; Zhao, Sanyuan et al.
Mobile Ad-hoc and Sensor Networks - 13th International Conference, MSN 2017, Revised Selected Papers. ed. / Liehuang Zhu; Sheng Zhong. Springer Verlag, 2018. p. 99-109 (Communications in Computer and Information Science; Vol. 747).

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

TY - GEN

T1 - Co-saliency detection based on siamese network

AU - Lei, Zhengchao

AU - Chai, Weiyan

AU - Zhao, Sanyuan

AU - Song, Hongmei

AU - Li, Fengxia

N1 - Publisher Copyright: © Springer Nature Singapore Pte Ltd. 2018.

PY - 2018

Y1 - 2018

N2 - Saliency detection in images attracts much research attention for its usage in numerous multimedia applications. Beside on the detection within the single image, co-saliency has been developed rapidly by detecting the same foreground objects in different images and trying to further promote the performance of object detection. This paper we propose a co-saliency detection method based on Siamese Network. By using Siamese Network, we get the similarity matrix of each image in superpixels. Guided by the single image saliency map, each saliency value, saliency score matrix is obtained to generate the multi image saliency map. Our final saliency map is a linear combination of these two saliency maps. The experiments show that our method performs better than other state-of-arts methods.

AB - Saliency detection in images attracts much research attention for its usage in numerous multimedia applications. Beside on the detection within the single image, co-saliency has been developed rapidly by detecting the same foreground objects in different images and trying to further promote the performance of object detection. This paper we propose a co-saliency detection method based on Siamese Network. By using Siamese Network, we get the similarity matrix of each image in superpixels. Guided by the single image saliency map, each saliency value, saliency score matrix is obtained to generate the multi image saliency map. Our final saliency map is a linear combination of these two saliency maps. The experiments show that our method performs better than other state-of-arts methods.

KW - Co-saliency detection

KW - Feature extraction

KW - Siamese network

UR - http://www.scopus.com/inward/record.url?scp=85045308766&partnerID=8YFLogxK

U2 - 10.1007/978-981-10-8890-2_8

DO - 10.1007/978-981-10-8890-2_8

M3 - Conference contribution

AN - SCOPUS:85045308766

SN - 9789811088896

T3 - Communications in Computer and Information Science

SP - 99

EP - 109

BT - Mobile Ad-hoc and Sensor Networks - 13th International Conference, MSN 2017, Revised Selected Papers

A2 - Zhu, Liehuang

A2 - Zhong, Sheng

PB - Springer Verlag

T2 - 13th International Conference on Mobile Ad-hoc and Sensor Networks, MSN 2017

Y2 - 17 December 2017 through 20 December 2017

ER -

Co-saliency detection based on siamese network

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this