CATS: co-saliency activated tracklet selection for video co-localization

Koteswar Rao Jerripothula; Jianfei Cai; Junsong Yuan

doi:10.1007/978-3-319-46478-7_12

CATS: co-saliency activated tracklet selection for video co-localization

Koteswar Rao Jerripothula, Jianfei Cai, Junsong Yuan

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

36 Citations (Scopus)

Abstract

Video co-localization is the task of jointly localizing common objects across videos. Due to the appearance variations both across the videos and within the video, it is a challenging problem to identify and track them without any supervision. In contrast to previous joint frameworks that use bounding box proposals to attack the problem, we propose to leverage co-saliency activated tracklets to address the challenge. To identify the common visual object, we first explore inter-video commonness, intra-video commonness, and motion saliency to generate the co-saliency maps. Object proposals of high objectness and co-saliency scores are tracked across short video intervals to build tracklets. The best tube for a video is obtained through tracklet selection from these intervals based on confidence and smoothness between the adjacent tracklets, with the help of dynamic programming. Experimental results on the benchmark YouTube Object dataset show that the proposed method outperforms state-of-the-art methods.

Original language	English
Title of host publication	Computer Vision – ECCV 2016
Subtitle of host publication	14th European Conference Amsterdam, The Netherlands, October 11–14, 2016 Proceedings, Part VII
Editors	Bastian Leibe, Jiri Matas, Nicu Sebe, Max Welling
Place of Publication	Cham Switzerland
Publisher	Springer
Pages	187-202
Number of pages	16
ISBN (Electronic)	9783319464787
ISBN (Print)	9783319464770
DOIs	https://doi.org/10.1007/978-3-319-46478-7_12
Publication status	Published - 2016
Externally published	Yes
Event	European Conference on Computer Vision 2016 - Amsterdam, Netherlands Duration: 11 Oct 2016 → 14 Oct 2016 Conference number: 14th http://www.eccv2016.org/ https://link.springer.com/book/10.1007/978-3-319-46448-0 (Proceedings)

Publication series

Name	Lecture Notes in Computer Science
Publisher	Springer
Volume	9911
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	European Conference on Computer Vision 2016
Abbreviated title	ECCV 2016
Country/Territory	Netherlands
City	Amsterdam
Period	11/10/16 → 14/10/16
Internet address	http://www.eccv2016.org/ https://link.springer.com/book/10.1007/978-3-319-46448-0 (Proceedings)

Keywords

Cats
Co-detection
Co-localization
Co-saliency
Tracklet
Video

Access to Document

10.1007/978-3-319-46478-7_12

Cite this

Jerripothula, K. R., Cai, J., & Yuan, J. (2016). CATS: co-saliency activated tracklet selection for video co-localization. In B. Leibe, J. Matas, N. Sebe, & M. Welling (Eds.), Computer Vision – ECCV 2016: 14th European Conference Amsterdam, The Netherlands, October 11–14, 2016 Proceedings, Part VII (pp. 187-202). (Lecture Notes in Computer Science ; Vol. 9911 ). Springer. https://doi.org/10.1007/978-3-319-46478-7_12

Jerripothula, Koteswar Rao ; Cai, Jianfei ; Yuan, Junsong. / CATS : co-saliency activated tracklet selection for video co-localization. Computer Vision – ECCV 2016: 14th European Conference Amsterdam, The Netherlands, October 11–14, 2016 Proceedings, Part VII. editor / Bastian Leibe ; Jiri Matas ; Nicu Sebe ; Max Welling. Cham Switzerland : Springer, 2016. pp. 187-202 (Lecture Notes in Computer Science ).

@inproceedings{d1284b59572f412bb9a90268c21004ef,

title = "CATS: co-saliency activated tracklet selection for video co-localization",

abstract = "Video co-localization is the task of jointly localizing common objects across videos. Due to the appearance variations both across the videos and within the video, it is a challenging problem to identify and track them without any supervision. In contrast to previous joint frameworks that use bounding box proposals to attack the problem, we propose to leverage co-saliency activated tracklets to address the challenge. To identify the common visual object, we first explore inter-video commonness, intra-video commonness, and motion saliency to generate the co-saliency maps. Object proposals of high objectness and co-saliency scores are tracked across short video intervals to build tracklets. The best tube for a video is obtained through tracklet selection from these intervals based on confidence and smoothness between the adjacent tracklets, with the help of dynamic programming. Experimental results on the benchmark YouTube Object dataset show that the proposed method outperforms state-of-the-art methods.",

keywords = "Cats, Co-detection, Co-localization, Co-saliency, Tracklet, Video",

author = "Jerripothula, {Koteswar Rao} and Jianfei Cai and Junsong Yuan",

year = "2016",

doi = "10.1007/978-3-319-46478-7_12",

language = "English",

isbn = "9783319464770",

series = "Lecture Notes in Computer Science ",

publisher = "Springer",

pages = "187--202",

editor = "Bastian Leibe and Jiri Matas and Nicu Sebe and Max Welling",

booktitle = "Computer Vision – ECCV 2016",

note = "European Conference on Computer Vision 2016, ECCV 2016 ; Conference date: 11-10-2016 Through 14-10-2016",

url = "http://www.eccv2016.org/, https://link.springer.com/book/10.1007/978-3-319-46448-0",

}

Jerripothula, KR, Cai, J & Yuan, J 2016, CATS: co-saliency activated tracklet selection for video co-localization. in B Leibe, J Matas, N Sebe & M Welling (eds), Computer Vision – ECCV 2016: 14th European Conference Amsterdam, The Netherlands, October 11–14, 2016 Proceedings, Part VII. Lecture Notes in Computer Science , vol. 9911 , Springer, Cham Switzerland, pp. 187-202, European Conference on Computer Vision 2016, Amsterdam, Netherlands, 11/10/16. https://doi.org/10.1007/978-3-319-46478-7_12

CATS: co-saliency activated tracklet selection for video co-localization. / Jerripothula, Koteswar Rao; Cai, Jianfei; Yuan, Junsong.
Computer Vision – ECCV 2016: 14th European Conference Amsterdam, The Netherlands, October 11–14, 2016 Proceedings, Part VII. ed. / Bastian Leibe; Jiri Matas; Nicu Sebe; Max Welling. Cham Switzerland: Springer, 2016. p. 187-202 (Lecture Notes in Computer Science ; Vol. 9911 ).

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

TY - GEN

T1 - CATS

T2 - European Conference on Computer Vision 2016

AU - Jerripothula, Koteswar Rao

AU - Cai, Jianfei

AU - Yuan, Junsong

N1 - Conference code: 14th

PY - 2016

Y1 - 2016

N2 - Video co-localization is the task of jointly localizing common objects across videos. Due to the appearance variations both across the videos and within the video, it is a challenging problem to identify and track them without any supervision. In contrast to previous joint frameworks that use bounding box proposals to attack the problem, we propose to leverage co-saliency activated tracklets to address the challenge. To identify the common visual object, we first explore inter-video commonness, intra-video commonness, and motion saliency to generate the co-saliency maps. Object proposals of high objectness and co-saliency scores are tracked across short video intervals to build tracklets. The best tube for a video is obtained through tracklet selection from these intervals based on confidence and smoothness between the adjacent tracklets, with the help of dynamic programming. Experimental results on the benchmark YouTube Object dataset show that the proposed method outperforms state-of-the-art methods.

AB - Video co-localization is the task of jointly localizing common objects across videos. Due to the appearance variations both across the videos and within the video, it is a challenging problem to identify and track them without any supervision. In contrast to previous joint frameworks that use bounding box proposals to attack the problem, we propose to leverage co-saliency activated tracklets to address the challenge. To identify the common visual object, we first explore inter-video commonness, intra-video commonness, and motion saliency to generate the co-saliency maps. Object proposals of high objectness and co-saliency scores are tracked across short video intervals to build tracklets. The best tube for a video is obtained through tracklet selection from these intervals based on confidence and smoothness between the adjacent tracklets, with the help of dynamic programming. Experimental results on the benchmark YouTube Object dataset show that the proposed method outperforms state-of-the-art methods.

KW - Cats

KW - Co-detection

KW - Co-localization

KW - Co-saliency

KW - Tracklet

KW - Video

UR - http://www.scopus.com/inward/record.url?scp=84990051025&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-46478-7_12

DO - 10.1007/978-3-319-46478-7_12

M3 - Conference Paper

AN - SCOPUS:84990051025

SN - 9783319464770

T3 - Lecture Notes in Computer Science

SP - 187

EP - 202

BT - Computer Vision – ECCV 2016

A2 - Leibe, Bastian

A2 - Matas, Jiri

A2 - Sebe, Nicu

A2 - Welling, Max

PB - Springer

CY - Cham Switzerland

Y2 - 11 October 2016 through 14 October 2016

ER -

Jerripothula KR, Cai J, Yuan J. CATS: co-saliency activated tracklet selection for video co-localization. In Leibe B, Matas J, Sebe N, Welling M, editors, Computer Vision – ECCV 2016: 14th European Conference Amsterdam, The Netherlands, October 11–14, 2016 Proceedings, Part VII. Cham Switzerland: Springer. 2016. p. 187-202. (Lecture Notes in Computer Science ). doi: 10.1007/978-3-319-46478-7_12

CATS: co-saliency activated tracklet selection for video co-localization

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Cite this