Image cosegmentation via multi-task learning

Qiang Zhang, Jiayu Zhou, Yilin Wang, Jieping Ye, Baoxin Li

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Citations (Scopus)

Abstract

Image segmentation has been studied in computer vision for many years and yet it remains a challenging task. One major difficulty arises from the diversity of the foreground, which often results in ambiguity of background-foreground separation, especially when prior knowledge is missing. To overcome this difficulty, cosegmentation methods were proposed, where a set of images sharing some common foreground objects are segmented simultaneously. Different models have been employed for exploring such a prior of common foreground. In this paper, we propose to formulate the image cosegmentaion problem using a multi-task learning framework, where segmentation of each image is viewed as one task and the prior of shared foreground is modeled via the intrinsic relatedness among the tasks. Compared with other existing methods, the proposed approach is able to simultaneously segment more than two images with relatively low computational cost. The proposed formulation, with three different embodiments, is evaluated on two benchmark datasets, the CMU iCoseg dataset and the MSRC dataset, with comparison to leading existing methods. Experimental results demonstrate the effectiveness of the proposed method.

Original languageEnglish (US)
Title of host publicationBMVC 2014 - Proceedings of the British Machine Vision Conference 2014
PublisherBritish Machine Vision Association, BMVA
StatePublished - 2014
Event25th British Machine Vision Conference, BMVC 2014 - Nottingham, United Kingdom
Duration: Sep 1 2014Sep 5 2014

Other

Other25th British Machine Vision Conference, BMVC 2014
CountryUnited Kingdom
CityNottingham
Period9/1/149/5/14

Fingerprint

Image segmentation
Computer vision
Costs

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition

Cite this

Zhang, Q., Zhou, J., Wang, Y., Ye, J., & Li, B. (2014). Image cosegmentation via multi-task learning. In BMVC 2014 - Proceedings of the British Machine Vision Conference 2014 British Machine Vision Association, BMVA.

Image cosegmentation via multi-task learning. / Zhang, Qiang; Zhou, Jiayu; Wang, Yilin; Ye, Jieping; Li, Baoxin.

BMVC 2014 - Proceedings of the British Machine Vision Conference 2014. British Machine Vision Association, BMVA, 2014.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Zhang, Q, Zhou, J, Wang, Y, Ye, J & Li, B 2014, Image cosegmentation via multi-task learning. in BMVC 2014 - Proceedings of the British Machine Vision Conference 2014. British Machine Vision Association, BMVA, 25th British Machine Vision Conference, BMVC 2014, Nottingham, United Kingdom, 9/1/14.
Zhang Q, Zhou J, Wang Y, Ye J, Li B. Image cosegmentation via multi-task learning. In BMVC 2014 - Proceedings of the British Machine Vision Conference 2014. British Machine Vision Association, BMVA. 2014
Zhang, Qiang ; Zhou, Jiayu ; Wang, Yilin ; Ye, Jieping ; Li, Baoxin. / Image cosegmentation via multi-task learning. BMVC 2014 - Proceedings of the British Machine Vision Conference 2014. British Machine Vision Association, BMVA, 2014.
@inproceedings{fc71b8d454dc4033882c286290812a72,
title = "Image cosegmentation via multi-task learning",
abstract = "Image segmentation has been studied in computer vision for many years and yet it remains a challenging task. One major difficulty arises from the diversity of the foreground, which often results in ambiguity of background-foreground separation, especially when prior knowledge is missing. To overcome this difficulty, cosegmentation methods were proposed, where a set of images sharing some common foreground objects are segmented simultaneously. Different models have been employed for exploring such a prior of common foreground. In this paper, we propose to formulate the image cosegmentaion problem using a multi-task learning framework, where segmentation of each image is viewed as one task and the prior of shared foreground is modeled via the intrinsic relatedness among the tasks. Compared with other existing methods, the proposed approach is able to simultaneously segment more than two images with relatively low computational cost. The proposed formulation, with three different embodiments, is evaluated on two benchmark datasets, the CMU iCoseg dataset and the MSRC dataset, with comparison to leading existing methods. Experimental results demonstrate the effectiveness of the proposed method.",
author = "Qiang Zhang and Jiayu Zhou and Yilin Wang and Jieping Ye and Baoxin Li",
year = "2014",
language = "English (US)",
booktitle = "BMVC 2014 - Proceedings of the British Machine Vision Conference 2014",
publisher = "British Machine Vision Association, BMVA",

}

TY - GEN

T1 - Image cosegmentation via multi-task learning

AU - Zhang, Qiang

AU - Zhou, Jiayu

AU - Wang, Yilin

AU - Ye, Jieping

AU - Li, Baoxin

PY - 2014

Y1 - 2014

N2 - Image segmentation has been studied in computer vision for many years and yet it remains a challenging task. One major difficulty arises from the diversity of the foreground, which often results in ambiguity of background-foreground separation, especially when prior knowledge is missing. To overcome this difficulty, cosegmentation methods were proposed, where a set of images sharing some common foreground objects are segmented simultaneously. Different models have been employed for exploring such a prior of common foreground. In this paper, we propose to formulate the image cosegmentaion problem using a multi-task learning framework, where segmentation of each image is viewed as one task and the prior of shared foreground is modeled via the intrinsic relatedness among the tasks. Compared with other existing methods, the proposed approach is able to simultaneously segment more than two images with relatively low computational cost. The proposed formulation, with three different embodiments, is evaluated on two benchmark datasets, the CMU iCoseg dataset and the MSRC dataset, with comparison to leading existing methods. Experimental results demonstrate the effectiveness of the proposed method.

AB - Image segmentation has been studied in computer vision for many years and yet it remains a challenging task. One major difficulty arises from the diversity of the foreground, which often results in ambiguity of background-foreground separation, especially when prior knowledge is missing. To overcome this difficulty, cosegmentation methods were proposed, where a set of images sharing some common foreground objects are segmented simultaneously. Different models have been employed for exploring such a prior of common foreground. In this paper, we propose to formulate the image cosegmentaion problem using a multi-task learning framework, where segmentation of each image is viewed as one task and the prior of shared foreground is modeled via the intrinsic relatedness among the tasks. Compared with other existing methods, the proposed approach is able to simultaneously segment more than two images with relatively low computational cost. The proposed formulation, with three different embodiments, is evaluated on two benchmark datasets, the CMU iCoseg dataset and the MSRC dataset, with comparison to leading existing methods. Experimental results demonstrate the effectiveness of the proposed method.

UR - http://www.scopus.com/inward/record.url?scp=84919741172&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84919741172&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84919741172

BT - BMVC 2014 - Proceedings of the British Machine Vision Conference 2014

PB - British Machine Vision Association, BMVA

ER -