Real-time salient object detection based on fully convolutional networks

Guangyu Nie; Yinan Guo; Yue Liu; Yongtian Wang

doi:10.1007/978-981-10-7389-2_19

Real-time salient object detection based on fully convolutional networks

Guangyu Nie, Yinan Guo, Yue Liu^*, Yongtian Wang

^*Corresponding author for this work

School of Optics and Photonics

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

11 Citations (Scopus)

Abstract

Salient object detection allows to take into account the visual content of images. In this paper, we train a real-time saliency model based on fully convolutional network (FCN), and then combine the energy maps with Gaussian filters to generate a multi-resolution image. The proposed method has been tested both qualitatively and quantitatively, by considering a representative set of ground truth images labeled with corresponding salient objects. Experimental results demonstrate that the proposed deep model significantly is superior to the-state-of-the-art approaches.

Original language	English
Title of host publication	Advances in Image and Graphics Technologies - 12th Chinese conference, IGTA 2017, Revised Selected Papers
Editors	Xiaoru Yuan, Henry Been-Lirn Duh, Yongtian Wang, Yue Liu, Jian Yang, Shengjin Wang, Ran He
Publisher	Springer Verlag
Pages	189-198
Number of pages	10
ISBN (Print)	9789811073885
DOIs	https://doi.org/10.1007/978-981-10-7389-2_19
Publication status	Published - 2018
Event	12th Chinese conference on Advances in Image and Graphics Technologies, IGTA 2017 - Beijing, China Duration: 30 Jun 2017 → 1 Jul 2017

Publication series

Name	Communications in Computer and Information Science
Volume	757
ISSN (Print)	1865-0929

Conference

Conference	12th Chinese conference on Advances in Image and Graphics Technologies, IGTA 2017
Country/Territory	China
City	Beijing
Period	30/06/17 → 1/07/17

Keywords

FCN
Real-time
Saliency detection

Access to Document

10.1007/978-981-10-7389-2_19

Cite this

Nie, G., Guo, Y., Liu, Y., & Wang, Y. (2018). Real-time salient object detection based on fully convolutional networks. In X. Yuan, H. B.-L. Duh, Y. Wang, Y. Liu, J. Yang, S. Wang, & R. He (Eds.), Advances in Image and Graphics Technologies - 12th Chinese conference, IGTA 2017, Revised Selected Papers (pp. 189-198). (Communications in Computer and Information Science; Vol. 757). Springer Verlag. https://doi.org/10.1007/978-981-10-7389-2_19

Nie, Guangyu ; Guo, Yinan ; Liu, Yue et al. / Real-time salient object detection based on fully convolutional networks. Advances in Image and Graphics Technologies - 12th Chinese conference, IGTA 2017, Revised Selected Papers. editor / Xiaoru Yuan ; Henry Been-Lirn Duh ; Yongtian Wang ; Yue Liu ; Jian Yang ; Shengjin Wang ; Ran He. Springer Verlag, 2018. pp. 189-198 (Communications in Computer and Information Science).

@inproceedings{5e5340ea55bf429085d731d9f81cb937,

title = "Real-time salient object detection based on fully convolutional networks",

abstract = "Salient object detection allows to take into account the visual content of images. In this paper, we train a real-time saliency model based on fully convolutional network (FCN), and then combine the energy maps with Gaussian filters to generate a multi-resolution image. The proposed method has been tested both qualitatively and quantitatively, by considering a representative set of ground truth images labeled with corresponding salient objects. Experimental results demonstrate that the proposed deep model significantly is superior to the-state-of-the-art approaches.",

keywords = "FCN, Real-time, Saliency detection",

author = "Guangyu Nie and Yinan Guo and Yue Liu and Yongtian Wang",

note = "Publisher Copyright: {\textcopyright} Springer Nature Singapore Pte Ltd. 2018.; 12th Chinese conference on Advances in Image and Graphics Technologies, IGTA 2017 ; Conference date: 30-06-2017 Through 01-07-2017",

year = "2018",

doi = "10.1007/978-981-10-7389-2_19",

language = "English",

isbn = "9789811073885",

series = "Communications in Computer and Information Science",

publisher = "Springer Verlag",

pages = "189--198",

editor = "Xiaoru Yuan and Duh, {Henry Been-Lirn} and Yongtian Wang and Yue Liu and Jian Yang and Shengjin Wang and Ran He",

booktitle = "Advances in Image and Graphics Technologies - 12th Chinese conference, IGTA 2017, Revised Selected Papers",

address = "Germany",

}

Nie, G, Guo, Y, Liu, Y & Wang, Y 2018, Real-time salient object detection based on fully convolutional networks. in X Yuan, HB-L Duh, Y Wang, Y Liu, J Yang, S Wang & R He (eds), Advances in Image and Graphics Technologies - 12th Chinese conference, IGTA 2017, Revised Selected Papers. Communications in Computer and Information Science, vol. 757, Springer Verlag, pp. 189-198, 12th Chinese conference on Advances in Image and Graphics Technologies, IGTA 2017, Beijing, China, 30/06/17. https://doi.org/10.1007/978-981-10-7389-2_19

Real-time salient object detection based on fully convolutional networks. / Nie, Guangyu; Guo, Yinan; Liu, Yue et al.
Advances in Image and Graphics Technologies - 12th Chinese conference, IGTA 2017, Revised Selected Papers. ed. / Xiaoru Yuan; Henry Been-Lirn Duh; Yongtian Wang; Yue Liu; Jian Yang; Shengjin Wang; Ran He. Springer Verlag, 2018. p. 189-198 (Communications in Computer and Information Science; Vol. 757).

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

TY - GEN

T1 - Real-time salient object detection based on fully convolutional networks

AU - Nie, Guangyu

AU - Guo, Yinan

AU - Liu, Yue

AU - Wang, Yongtian

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

PY - 2018

Y1 - 2018

N2 - Salient object detection allows to take into account the visual content of images. In this paper, we train a real-time saliency model based on fully convolutional network (FCN), and then combine the energy maps with Gaussian filters to generate a multi-resolution image. The proposed method has been tested both qualitatively and quantitatively, by considering a representative set of ground truth images labeled with corresponding salient objects. Experimental results demonstrate that the proposed deep model significantly is superior to the-state-of-the-art approaches.

AB - Salient object detection allows to take into account the visual content of images. In this paper, we train a real-time saliency model based on fully convolutional network (FCN), and then combine the energy maps with Gaussian filters to generate a multi-resolution image. The proposed method has been tested both qualitatively and quantitatively, by considering a representative set of ground truth images labeled with corresponding salient objects. Experimental results demonstrate that the proposed deep model significantly is superior to the-state-of-the-art approaches.

KW - FCN

KW - Real-time

KW - Saliency detection

UR - http://www.scopus.com/inward/record.url?scp=85036451646&partnerID=8YFLogxK

U2 - 10.1007/978-981-10-7389-2_19

DO - 10.1007/978-981-10-7389-2_19

M3 - Conference contribution

AN - SCOPUS:85036451646

SN - 9789811073885

T3 - Communications in Computer and Information Science

SP - 189

EP - 198

BT - Advances in Image and Graphics Technologies - 12th Chinese conference, IGTA 2017, Revised Selected Papers

A2 - Yuan, Xiaoru

A2 - Duh, Henry Been-Lirn

A2 - Wang, Yongtian

A2 - Liu, Yue

A2 - Yang, Jian

A2 - Wang, Shengjin

A2 - He, Ran

PB - Springer Verlag

T2 - 12th Chinese conference on Advances in Image and Graphics Technologies, IGTA 2017

Y2 - 30 June 2017 through 1 July 2017

ER -

Nie G, Guo Y, Liu Y , Wang Y. Real-time salient object detection based on fully convolutional networks. In Yuan X, Duh HBL, Wang Y, Liu Y, Yang J, Wang S, He R, editors, Advances in Image and Graphics Technologies - 12th Chinese conference, IGTA 2017, Revised Selected Papers. Springer Verlag. 2018. p. 189-198. (Communications in Computer and Information Science). doi: 10.1007/978-981-10-7389-2_19

Real-time salient object detection based on fully convolutional networks

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this