Video saliency detection based on temporal difference and pixel gradient

Xiangwei Lu; Muwei Jian; Rui Wang; Zhichao Yun; Peiguang Lin; Hui Yu

doi:10.23919/ICAC50006.2021.9594136

Video saliency detection based on temporal difference and pixel gradient

Xiangwei Lu, Muwei Jian, Rui Wang, Zhichao Yun^*, Peiguang Lin, Hui Yu

^*Corresponding author for this work

School of Creative Technologies

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

27 Downloads (Pure)

Abstract

Even though temporal information matters for the quality of video saliency detection, many problems such as bad performance in time-space coherence and edge continuity still face present network frameworks. In response to these problems, this paper designs a full convolutional neural network, which integrates temporal differential and pixel gradient to fine tune the edges of targets. Meanwhile, the changes of pixel gradients of original images are used to recursively improve the continuity of target edges and details of central areas. The method presented in the paper has been tested with two available public datasets and its effectiveness been proved after it being compared with 6 other widely accepted methods.

Original language	English
Title of host publication	2021 26th International Conference on Automation and Computing
Subtitle of host publication	System Intelligence through Automation and Computing, ICAC 2021
Editors	Chenguang Yang
Publisher	Institute of Electrical and Electronics Engineers Inc.
Number of pages	5
ISBN (Electronic)	9781860435577
ISBN (Print)	9781665443524
DOIs	https://doi.org/10.23919/ICAC50006.2021.9594136
Publication status	Early online - 15 Nov 2021
Event	26th International Conference on Automation and Computing - Portsmouth, United Kingdom Duration: 2 Sept 2021 → 4 Sept 2021

Conference

Conference	26th International Conference on Automation and Computing
Abbreviated title	ICAC 2021
Country/Territory	United Kingdom
City	Portsmouth
Period	2/09/21 → 4/09/21

Keywords

Co-Attention
Edge refinement
Pixels gradient
Temporal difference
Video saliency detection

Access to Document

10.23919/ICAC50006.2021.9594136

Video_saliency_detection_based_on_temporal_difference_and_pixel_gradient_pp
© 2021 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
Accepted author manuscript (Post-print), 734 KB

Cite this

Lu, X., Jian, M., Wang, R., Yun, Z., Lin, P., & Yu, H. (2021). Video saliency detection based on temporal difference and pixel gradient. In C. Yang (Ed.), 2021 26th International Conference on Automation and Computing: System Intelligence through Automation and Computing, ICAC 2021 Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.23919/ICAC50006.2021.9594136

@inproceedings{f8851f0cd0fb463f8a6eff70062094f7,

title = "Video saliency detection based on temporal difference and pixel gradient",

abstract = "Even though temporal information matters for the quality of video saliency detection, many problems such as bad performance in time-space coherence and edge continuity still face present network frameworks. In response to these problems, this paper designs a full convolutional neural network, which integrates temporal differential and pixel gradient to fine tune the edges of targets. Meanwhile, the changes of pixel gradients of original images are used to recursively improve the continuity of target edges and details of central areas. The method presented in the paper has been tested with two available public datasets and its effectiveness been proved after it being compared with 6 other widely accepted methods. ",

keywords = "Co-Attention, Edge refinement, Pixels gradient, Temporal difference, Video saliency detection",

author = "Xiangwei Lu and Muwei Jian and Rui Wang and Zhichao Yun and Peiguang Lin and Hui Yu",

note = "No ISSN; 26th International Conference on Automation and Computing, ICAC 2021 ; Conference date: 02-09-2021 Through 04-09-2021",

year = "2021",

month = nov,

day = "15",

doi = "10.23919/ICAC50006.2021.9594136",

language = "English",

isbn = "9781665443524",

editor = "Chenguang Yang",

booktitle = "2021 26th International Conference on Automation and Computing",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

address = "United States",

}

Lu, X, Jian, M, Wang, R, Yun, Z, Lin, P & Yu, H 2021, Video saliency detection based on temporal difference and pixel gradient. in C Yang (ed.), 2021 26th International Conference on Automation and Computing: System Intelligence through Automation and Computing, ICAC 2021. Institute of Electrical and Electronics Engineers Inc., 26th International Conference on Automation and Computing, Portsmouth, United Kingdom, 2/09/21. https://doi.org/10.23919/ICAC50006.2021.9594136

Video saliency detection based on temporal difference and pixel gradient. / Lu, Xiangwei; Jian, Muwei; Wang, Rui et al.
2021 26th International Conference on Automation and Computing: System Intelligence through Automation and Computing, ICAC 2021. ed. / Chenguang Yang. Institute of Electrical and Electronics Engineers Inc., 2021.

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

TY - GEN

T1 - Video saliency detection based on temporal difference and pixel gradient

AU - Lu, Xiangwei

AU - Jian, Muwei

AU - Wang, Rui

AU - Yun, Zhichao

AU - Lin, Peiguang

AU - Yu, Hui

N1 - No ISSN

PY - 2021/11/15

Y1 - 2021/11/15

N2 - Even though temporal information matters for the quality of video saliency detection, many problems such as bad performance in time-space coherence and edge continuity still face present network frameworks. In response to these problems, this paper designs a full convolutional neural network, which integrates temporal differential and pixel gradient to fine tune the edges of targets. Meanwhile, the changes of pixel gradients of original images are used to recursively improve the continuity of target edges and details of central areas. The method presented in the paper has been tested with two available public datasets and its effectiveness been proved after it being compared with 6 other widely accepted methods.

AB - Even though temporal information matters for the quality of video saliency detection, many problems such as bad performance in time-space coherence and edge continuity still face present network frameworks. In response to these problems, this paper designs a full convolutional neural network, which integrates temporal differential and pixel gradient to fine tune the edges of targets. Meanwhile, the changes of pixel gradients of original images are used to recursively improve the continuity of target edges and details of central areas. The method presented in the paper has been tested with two available public datasets and its effectiveness been proved after it being compared with 6 other widely accepted methods.

KW - Co-Attention

KW - Edge refinement

KW - Pixels gradient

KW - Temporal difference

KW - Video saliency detection

U2 - 10.23919/ICAC50006.2021.9594136

DO - 10.23919/ICAC50006.2021.9594136

M3 - Conference contribution

AN - SCOPUS:85123219528

SN - 9781665443524

BT - 2021 26th International Conference on Automation and Computing

A2 - Yang, Chenguang

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 26th International Conference on Automation and Computing

Y2 - 2 September 2021 through 4 September 2021

ER -

Lu X, Jian M, Wang R, Yun Z, Lin P, Yu H. Video saliency detection based on temporal difference and pixel gradient. In Yang C, editor, 2021 26th International Conference on Automation and Computing: System Intelligence through Automation and Computing, ICAC 2021. Institute of Electrical and Electronics Engineers Inc. 2021 Epub 2021 Nov 15. doi: 10.23919/ICAC50006.2021.9594136

Video saliency detection based on temporal difference and pixel gradient

Abstract

Conference

Keywords

Access to Document

Fingerprint

Cite this