Deep Verification Learning

Fieke Hillerström; Raymond Veldhuis; Luuk Spreeuwers

Deep Verification Learning

Fieke Hillerström, Raymond Veldhuis, Luuk Spreeuwers

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

59 Downloads (Pure)

Abstract

Deep learning for biometrics has increasingly gained attention over the last years.
The expansion of computational power and the increasing dataset sizes, increased
verification performances. However, large datasets are not available for every ap-
plication. We introduce Deep Verification Learning, to reduce network complex-
ity and train on smaller datasets. Deep Verification Learning takes two images
to be verified at the input of a network, and trains directly towards a verification
score. We applied Deep Verification Learning on the face verification task, also
it could be extended to other biometric modalities.

Original language	English
Title of host publication	Proceedings of the 2017 Symposium on Information Theory and Signal Processing in the Benelux
Subtitle of host publication	May 11-12, 2017, Delft University of Technology, Delft, the Netherlands
Editors	Richard Heusden, Jos H. Weber
Place of Publication	Delft
Publisher	Delft University of Technology
Pages	97-104
ISBN (Print)	978-94-6186-811-4
Publication status	Published - 2017
Event	38th WIC Symposium on Information Theory in the Benelux 2017 - Delft, Netherlands Duration: 11 May 2017 → 12 May 2017 Conference number: 38

Conference

Conference	38th WIC Symposium on Information Theory in the Benelux 2017
Country/Territory	Netherlands
City	Delft
Period	11/05/17 → 12/05/17

Keywords

SCS-Safety
FISWG
Eyebrows
Human verification performance

Access to Document

Hillerström2017deepFinal published version, 1.4 MB

Cite this

@inproceedings{299b7059c841472fb0f047ddbc19dd40,

title = "Deep Verification Learning",

abstract = "Deep learning for biometrics has increasingly gained attention over the last years.The expansion of computational power and the increasing dataset sizes, increasedverification performances. However, large datasets are not available for every ap-plication. We introduce Deep Verification Learning, to reduce network complex-ity and train on smaller datasets. Deep Verification Learning takes two imagesto be verified at the input of a network, and trains directly towards a verificationscore. We applied Deep Verification Learning on the face verification task, alsoit could be extended to other biometric modalities.",

keywords = "SCS-Safety, FISWG, Eyebrows, Human verification performance",

author = "Fieke Hillerstr{\"o}m and Raymond Veldhuis and Luuk Spreeuwers",

year = "2017",

language = "English",

isbn = "978-94-6186-811-4",

pages = "97--104",

editor = "Richard Heusden and Weber, {Jos H.}",

booktitle = "Proceedings of the 2017 Symposium on Information Theory and Signal Processing in the Benelux",

publisher = "Delft University of Technology",

address = "Netherlands",

note = "38th WIC Symposium on Information Theory in the Benelux 2017 ; Conference date: 11-05-2017 Through 12-05-2017",

}

Hillerström, F, Veldhuis, R & Spreeuwers, L 2017, Deep Verification Learning. in R Heusden & JH Weber (eds), Proceedings of the 2017 Symposium on Information Theory and Signal Processing in the Benelux: May 11-12, 2017, Delft University of Technology, Delft, the Netherlands. Delft University of Technology, Delft, pp. 97-104, 38th WIC Symposium on Information Theory in the Benelux 2017, Delft, Netherlands, 11/05/17.

Deep Verification Learning. / Hillerström, Fieke; Veldhuis, Raymond ; Spreeuwers, Luuk.
Proceedings of the 2017 Symposium on Information Theory and Signal Processing in the Benelux: May 11-12, 2017, Delft University of Technology, Delft, the Netherlands. ed. / Richard Heusden; Jos H. Weber. Delft: Delft University of Technology, 2017. p. 97-104.

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

TY - GEN

T1 - Deep Verification Learning

AU - Hillerström, Fieke

AU - Veldhuis, Raymond

AU - Spreeuwers, Luuk

N1 - Conference code: 38

PY - 2017

Y1 - 2017

N2 - Deep learning for biometrics has increasingly gained attention over the last years.The expansion of computational power and the increasing dataset sizes, increasedverification performances. However, large datasets are not available for every ap-plication. We introduce Deep Verification Learning, to reduce network complex-ity and train on smaller datasets. Deep Verification Learning takes two imagesto be verified at the input of a network, and trains directly towards a verificationscore. We applied Deep Verification Learning on the face verification task, alsoit could be extended to other biometric modalities.

AB - Deep learning for biometrics has increasingly gained attention over the last years.The expansion of computational power and the increasing dataset sizes, increasedverification performances. However, large datasets are not available for every ap-plication. We introduce Deep Verification Learning, to reduce network complex-ity and train on smaller datasets. Deep Verification Learning takes two imagesto be verified at the input of a network, and trains directly towards a verificationscore. We applied Deep Verification Learning on the face verification task, alsoit could be extended to other biometric modalities.

KW - SCS-Safety

KW - FISWG

KW - Eyebrows

KW - Human verification performance

M3 - Conference contribution

SN - 978-94-6186-811-4

SP - 97

EP - 104

BT - Proceedings of the 2017 Symposium on Information Theory and Signal Processing in the Benelux

A2 - Heusden, Richard

A2 - Weber, Jos H.

PB - Delft University of Technology

CY - Delft

T2 - 38th WIC Symposium on Information Theory in the Benelux 2017

Y2 - 11 May 2017 through 12 May 2017

ER -

Deep Verification Learning

Abstract

Conference

Keywords

Access to Document

Fingerprint

Cite this