Category independent object proposals

Ian Endres; Derek Hoiem

doi:10.1007/978-3-642-15555-0_42

Category independent object proposals

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

Abstract

We propose a category-independent method to produce a bag of regions and rank them, such that top-ranked regions are likely to be good segmentations of different objects. Our key objectives are completeness and diversity: every object should have at least one good proposed region, and a diverse set should be top-ranked. Our approach is to generate a set of segmentations by performing graph cuts based on a seed region and a learned affinity function. Then, the regions are ranked using structured learning based on various cues. Our experiments on BSDS and PASCAL VOC 2008 demonstrate our ability to find most objects within a small bag of proposed regions.

Original language	English (US)
Title of host publication	Computer Vision, ECCV 2010 - 11th European Conference on Computer Vision, Proceedings
Publisher	Springer
Pages	575-588
Number of pages	14
Edition	PART 5
ISBN (Print)	3642155545, 9783642155543
DOIs	https://doi.org/10.1007/978-3-642-15555-0_42
State	Published - 2010
Event	11th European Conference on Computer Vision, ECCV 2010 - Heraklion, Crete, Greece Duration: Sep 10 2010 → Sep 11 2010

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Number	PART 5
Volume	6315 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	11th European Conference on Computer Vision, ECCV 2010
Country/Territory	Greece
City	Heraklion, Crete
Period	9/10/10 → 9/11/10

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Online availability

10.1007/978-3-642-15555-0_42

Library availability

Discover UIUC Full Text

Cite this

Endres, I., & Hoiem, D. (2010). Category independent object proposals. In Computer Vision, ECCV 2010 - 11th European Conference on Computer Vision, Proceedings (PART 5 ed., pp. 575-588). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6315 LNCS, No. PART 5). Springer. https://doi.org/10.1007/978-3-642-15555-0_42

Category independent object proposals. / Endres, Ian; Hoiem, Derek.
Computer Vision, ECCV 2010 - 11th European Conference on Computer Vision, Proceedings. PART 5. ed. Springer, 2010. p. 575-588 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6315 LNCS, No. PART 5).

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

Endres, I & Hoiem, D 2010, Category independent object proposals. in Computer Vision, ECCV 2010 - 11th European Conference on Computer Vision, Proceedings. PART 5 edn, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 5, vol. 6315 LNCS, Springer, pp. 575-588, 11th European Conference on Computer Vision, ECCV 2010, Heraklion, Crete, Greece, 9/10/10. https://doi.org/10.1007/978-3-642-15555-0_42

@inproceedings{acb3a46d7ece43488bfca36adf774fe1,

title = "Category independent object proposals",

abstract = "We propose a category-independent method to produce a bag of regions and rank them, such that top-ranked regions are likely to be good segmentations of different objects. Our key objectives are completeness and diversity: every object should have at least one good proposed region, and a diverse set should be top-ranked. Our approach is to generate a set of segmentations by performing graph cuts based on a seed region and a learned affinity function. Then, the regions are ranked using structured learning based on various cues. Our experiments on BSDS and PASCAL VOC 2008 demonstrate our ability to find most objects within a small bag of proposed regions.",

author = "Ian Endres and Derek Hoiem",

year = "2010",

doi = "10.1007/978-3-642-15555-0_42",

language = "English (US)",

isbn = "3642155545",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer",

number = "PART 5",

pages = "575--588",

booktitle = "Computer Vision, ECCV 2010 - 11th European Conference on Computer Vision, Proceedings",

address = "Germany",

edition = "PART 5",

note = "11th European Conference on Computer Vision, ECCV 2010 ; Conference date: 10-09-2010 Through 11-09-2010",

}

TY - GEN

T1 - Category independent object proposals

AU - Endres, Ian

AU - Hoiem, Derek

PY - 2010

Y1 - 2010

N2 - We propose a category-independent method to produce a bag of regions and rank them, such that top-ranked regions are likely to be good segmentations of different objects. Our key objectives are completeness and diversity: every object should have at least one good proposed region, and a diverse set should be top-ranked. Our approach is to generate a set of segmentations by performing graph cuts based on a seed region and a learned affinity function. Then, the regions are ranked using structured learning based on various cues. Our experiments on BSDS and PASCAL VOC 2008 demonstrate our ability to find most objects within a small bag of proposed regions.

AB - We propose a category-independent method to produce a bag of regions and rank them, such that top-ranked regions are likely to be good segmentations of different objects. Our key objectives are completeness and diversity: every object should have at least one good proposed region, and a diverse set should be top-ranked. Our approach is to generate a set of segmentations by performing graph cuts based on a seed region and a learned affinity function. Then, the regions are ranked using structured learning based on various cues. Our experiments on BSDS and PASCAL VOC 2008 demonstrate our ability to find most objects within a small bag of proposed regions.

UR - http://www.scopus.com/inward/record.url?scp=78149308041&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78149308041&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-15555-0_42

DO - 10.1007/978-3-642-15555-0_42

M3 - Conference contribution

AN - SCOPUS:78149308041

SN - 3642155545

SN - 9783642155543

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 575

EP - 588

BT - Computer Vision, ECCV 2010 - 11th European Conference on Computer Vision, Proceedings

PB - Springer

T2 - 11th European Conference on Computer Vision, ECCV 2010

Y2 - 10 September 2010 through 11 September 2010

ER -

Category independent object proposals

Abstract

Publication series

Conference

ASJC Scopus subject areas

Online availability

Library availability

Related links

Fingerprint

Cite this