Hyperbolic Image Segmentation

Mina Ghadimi Atigh; Julian Schoep; Erman Acar; Nanne Van Noord; Pascal Mettes

doi:10.1109/CVPR52688.2022.00441

Hyperbolic Image Segmentation

Mina Ghadimi Atigh, Julian Schoep, Erman Acar, Nanne Van Noord, Pascal Mettes

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

Abstract

For image segmentation, the current standard is to perform pixel-level optimization and inference in Euclidean output embedding spaces through linear hyperplanes. In this work, we show that hyperbolic manifolds provide a valuable alternative for image segmentation and propose a tractable formulation of hierarchical pixel-level classification in hyperbolic space. Hyperbolic Image Segmentation opens up new possibilities and practical benefits for segmentation, such as uncertainty estimation and boundary information for free, zero-label generalization, and increased performance in low-dimensional output embeddings.

Original language	English
Title of host publication	2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
Subtitle of host publication	[Proceedings]
Publisher	IEEE Computer Society
Pages	4443-4452
Number of pages	10
ISBN (Electronic)	9781665469463
DOIs	https://doi.org/10.1109/CVPR52688.2022.00441
Publication status	Published - 27 Sept 2022
Event	2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022 - New Orleans, United States Duration: 19 Jun 2022 → 24 Jun 2022

Publication series

Name	Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
Volume	2022-June
ISSN (Print)	1063-6919

Conference

Conference	2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022
Country/Territory	United States
City	New Orleans
Period	19/06/22 → 24/06/22

Bibliographical note

Funding Information:
Acknowledgments. Erman Acar is generously funded by the Hybrid Intelligence Project which is financed by the Dutch Ministry of Education, Culture and Science with project number 024.004.022.

Publisher Copyright:
© 2022 IEEE.

Funding

Acknowledgments. Erman Acar is generously funded by the Hybrid Intelligence Project which is financed by the Dutch Ministry of Education, Culture and Science with project number 024.004.022.

Funders	Funder number
Ministerie van Onderwijs, Cultuur en Wetenschap
Not added	024.004.022

Keywords

grouping and shape analysis
Segmentation

Access to Document

10.1109/CVPR52688.2022.00441

Persistent URL (handle)

Persistent link

Cite this

Atigh, M. G., Schoep, J., Acar, E., Van Noord, N., & Mettes, P. (2022). Hyperbolic Image Segmentation. In 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR): [Proceedings] (pp. 4443-4452). (Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition; Vol. 2022-June). IEEE Computer Society. https://doi.org/10.1109/CVPR52688.2022.00441

@inproceedings{0cc6a398741d4228b2e88949d918ac9b,

title = "Hyperbolic Image Segmentation",

abstract = "For image segmentation, the current standard is to perform pixel-level optimization and inference in Euclidean output embedding spaces through linear hyperplanes. In this work, we show that hyperbolic manifolds provide a valuable alternative for image segmentation and propose a tractable formulation of hierarchical pixel-level classification in hyperbolic space. Hyperbolic Image Segmentation opens up new possibilities and practical benefits for segmentation, such as uncertainty estimation and boundary information for free, zero-label generalization, and increased performance in low-dimensional output embeddings.",

keywords = "grouping and shape analysis, Segmentation",

author = "Atigh, {Mina Ghadimi} and Julian Schoep and Erman Acar and {Van Noord}, Nanne and Pascal Mettes",

note = "Funding Information: Acknowledgments. Erman Acar is generously funded by the Hybrid Intelligence Project which is financed by the Dutch Ministry of Education, Culture and Science with project number 024.004.022. Publisher Copyright: {\textcopyright} 2022 IEEE.; 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022 ; Conference date: 19-06-2022 Through 24-06-2022",

year = "2022",

month = sep,

day = "27",

doi = "10.1109/CVPR52688.2022.00441",

language = "English",

series = "Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition",

publisher = "IEEE Computer Society",

pages = "4443--4452",

booktitle = "2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)",

address = "United States",

}

Atigh, MG, Schoep, J, Acar, E, Van Noord, N & Mettes, P 2022, Hyperbolic Image Segmentation. in 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR): [Proceedings]. Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol. 2022-June, IEEE Computer Society, pp. 4443-4452, 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022, New Orleans, United States, 19/06/22. https://doi.org/10.1109/CVPR52688.2022.00441

Hyperbolic Image Segmentation. / Atigh, Mina Ghadimi; Schoep, Julian; Acar, Erman et al.
2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR): [Proceedings]. IEEE Computer Society, 2022. p. 4443-4452 (Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition; Vol. 2022-June).

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

TY - GEN

T1 - Hyperbolic Image Segmentation

AU - Atigh, Mina Ghadimi

AU - Schoep, Julian

AU - Acar, Erman

AU - Van Noord, Nanne

AU - Mettes, Pascal

N1 - Funding Information: Acknowledgments. Erman Acar is generously funded by the Hybrid Intelligence Project which is financed by the Dutch Ministry of Education, Culture and Science with project number 024.004.022. Publisher Copyright: © 2022 IEEE.

PY - 2022/9/27

Y1 - 2022/9/27

N2 - For image segmentation, the current standard is to perform pixel-level optimization and inference in Euclidean output embedding spaces through linear hyperplanes. In this work, we show that hyperbolic manifolds provide a valuable alternative for image segmentation and propose a tractable formulation of hierarchical pixel-level classification in hyperbolic space. Hyperbolic Image Segmentation opens up new possibilities and practical benefits for segmentation, such as uncertainty estimation and boundary information for free, zero-label generalization, and increased performance in low-dimensional output embeddings.

AB - For image segmentation, the current standard is to perform pixel-level optimization and inference in Euclidean output embedding spaces through linear hyperplanes. In this work, we show that hyperbolic manifolds provide a valuable alternative for image segmentation and propose a tractable formulation of hierarchical pixel-level classification in hyperbolic space. Hyperbolic Image Segmentation opens up new possibilities and practical benefits for segmentation, such as uncertainty estimation and boundary information for free, zero-label generalization, and increased performance in low-dimensional output embeddings.

KW - grouping and shape analysis

KW - Segmentation

UR - http://www.scopus.com/inward/record.url?scp=85143157557&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85143157557&partnerID=8YFLogxK

U2 - 10.1109/CVPR52688.2022.00441

DO - 10.1109/CVPR52688.2022.00441

M3 - Conference contribution

AN - SCOPUS:85143157557

T3 - Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition

SP - 4443

EP - 4452

BT - 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

PB - IEEE Computer Society

T2 - 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022

Y2 - 19 June 2022 through 24 June 2022

ER -

Hyperbolic Image Segmentation

Abstract

Publication series

Conference

Bibliographical note

Funding

Keywords

Access to Document

Persistent URL (handle)

Other files and links

Fingerprint

Cite this