Estimating relative depth in single images via rankboost

Publikation: Beitrag in Buch/Bericht/Sammelwerk/Konferenzband › Aufsatz in Konferenzband › Forschung › Peer-Review

Autoren

  • Ralph Ewerth
  • Matthias Springstein
  • Alexander Balz
  • Jan Gehlhaar
  • Tolga Naziyok
  • Krzysztof Dembczynski
  • Eyke Hullermeier
  • Eric Müller-Budack

Organisationseinheiten

Externe Organisationen

  • Technische Informationsbibliothek (TIB) Leibniz-Informationszentrum Technik und Naturwissenschaften und Universitätsbibliothek
  • Philipps-Universität Marburg
  • Uniwersytet Ekonomiczny w Poznaniu
  • Universität Paderborn
Forschungs-netzwerk anzeigen

Details

OriginalspracheEnglisch
Titel des Sammelwerks2017 IEEE International Conference on Multimedia and Expo, ICME 2017
Seiten919-924
Seitenumfang6
ISBN (elektronisch)9781509060672
PublikationsstatusVeröffentlicht - 28 Aug. 2017
Veranstaltung2017 IEEE International Conference on Multimedia and Expo, ICME 2017 - Hong Kong, Hongkong
Dauer: 10 Juli 2017 → 14 Juli 2017

Abstract

In this paper, we present a novel approach to estimate the relative depth of regions in monocular images. There are several contributions. First, the task of monocular depth estimation is considered as a learning-to-rank problem which offers several advantages compared to regression approaches. Second, monocular depth clues of human perception are modeled in a systematic manner. Third, we show that these depth clues can be modeled and integrated appropriately in a Rankboost framework. For this purpose, a space-efficient version of Rankboost is derived that makes it applicable to rank a large number of objects, as posed by the given problem. Finally, the monocular depth clues are combined with results from a deep learning approach. Experimental results show that the error rate is reduced by adding the monocular features while outperforming state-of-the-art systems.

ASJC Scopus Sachgebiete

Zitieren

Estimating relative depth in single images via rankboost. / Ewerth, Ralph; Springstein, Matthias; Balz, Alexander et al.
2017 IEEE International Conference on Multimedia and Expo, ICME 2017. 2017. S. 919-924 8019434.

Publikation: Beitrag in Buch/Bericht/Sammelwerk/Konferenzband › Aufsatz in Konferenzband › Forschung › Peer-Review

Ewerth, R, Springstein, M, Balz, A, Gehlhaar, J, Naziyok, T, Dembczynski, K, Hullermeier, E & Müller-Budack, E 2017, Estimating relative depth in single images via rankboost. in 2017 IEEE International Conference on Multimedia and Expo, ICME 2017., 8019434, S. 919-924, 2017 IEEE International Conference on Multimedia and Expo, ICME 2017, Hong Kong, Hongkong, 10 Juli 2017. https://doi.org/10.1109/ICME.2017.8019434
Ewerth, R., Springstein, M., Balz, A., Gehlhaar, J., Naziyok, T., Dembczynski, K., Hullermeier, E., & Müller-Budack, E. (2017). Estimating relative depth in single images via rankboost. In 2017 IEEE International Conference on Multimedia and Expo, ICME 2017 (S. 919-924). Artikel 8019434 https://doi.org/10.1109/ICME.2017.8019434
Ewerth R, Springstein M, Balz A, Gehlhaar J, Naziyok T, Dembczynski K et al. Estimating relative depth in single images via rankboost. in 2017 IEEE International Conference on Multimedia and Expo, ICME 2017. 2017. S. 919-924. 8019434 doi: 10.1109/ICME.2017.8019434
Ewerth, Ralph ; Springstein, Matthias ; Balz, Alexander et al. / Estimating relative depth in single images via rankboost. 2017 IEEE International Conference on Multimedia and Expo, ICME 2017. 2017. S. 919-924
Download
@inproceedings{8f276160d23945dcb4b015e711fbe696,
title = "Estimating relative depth in single images via rankboost",
abstract = "In this paper, we present a novel approach to estimate the relative depth of regions in monocular images. There are several contributions. First, the task of monocular depth estimation is considered as a learning-to-rank problem which offers several advantages compared to regression approaches. Second, monocular depth clues of human perception are modeled in a systematic manner. Third, we show that these depth clues can be modeled and integrated appropriately in a Rankboost framework. For this purpose, a space-efficient version of Rankboost is derived that makes it applicable to rank a large number of objects, as posed by the given problem. Finally, the monocular depth clues are combined with results from a deep learning approach. Experimental results show that the error rate is reduced by adding the monocular features while outperforming state-of-the-art systems.",
keywords = "Monocular depth estimation, Rankboost",
author = "Ralph Ewerth and Matthias Springstein and Alexander Balz and Jan Gehlhaar and Tolga Naziyok and Krzysztof Dembczynski and Eyke Hullermeier and Eric M{\"u}ller-Budack",
note = "Publisher Copyright: {\textcopyright} 2017 IEEE. Copyright: Copyright 2017 Elsevier B.V., All rights reserved.; 2017 IEEE International Conference on Multimedia and Expo, ICME 2017 ; Conference date: 10-07-2017 Through 14-07-2017",
year = "2017",
month = aug,
day = "28",
doi = "10.1109/ICME.2017.8019434",
language = "English",
pages = "919--924",
booktitle = "2017 IEEE International Conference on Multimedia and Expo, ICME 2017",

}

Download

TY - GEN

T1 - Estimating relative depth in single images via rankboost

AU - Ewerth, Ralph

AU - Springstein, Matthias

AU - Balz, Alexander

AU - Gehlhaar, Jan

AU - Naziyok, Tolga

AU - Dembczynski, Krzysztof

AU - Hullermeier, Eyke

AU - Müller-Budack, Eric

N1 - Publisher Copyright: © 2017 IEEE. Copyright: Copyright 2017 Elsevier B.V., All rights reserved.

PY - 2017/8/28

Y1 - 2017/8/28

N2 - In this paper, we present a novel approach to estimate the relative depth of regions in monocular images. There are several contributions. First, the task of monocular depth estimation is considered as a learning-to-rank problem which offers several advantages compared to regression approaches. Second, monocular depth clues of human perception are modeled in a systematic manner. Third, we show that these depth clues can be modeled and integrated appropriately in a Rankboost framework. For this purpose, a space-efficient version of Rankboost is derived that makes it applicable to rank a large number of objects, as posed by the given problem. Finally, the monocular depth clues are combined with results from a deep learning approach. Experimental results show that the error rate is reduced by adding the monocular features while outperforming state-of-the-art systems.

AB - In this paper, we present a novel approach to estimate the relative depth of regions in monocular images. There are several contributions. First, the task of monocular depth estimation is considered as a learning-to-rank problem which offers several advantages compared to regression approaches. Second, monocular depth clues of human perception are modeled in a systematic manner. Third, we show that these depth clues can be modeled and integrated appropriately in a Rankboost framework. For this purpose, a space-efficient version of Rankboost is derived that makes it applicable to rank a large number of objects, as posed by the given problem. Finally, the monocular depth clues are combined with results from a deep learning approach. Experimental results show that the error rate is reduced by adding the monocular features while outperforming state-of-the-art systems.

KW - Monocular depth estimation

KW - Rankboost

UR - http://www.scopus.com/inward/record.url?scp=85030229368&partnerID=8YFLogxK

U2 - 10.1109/ICME.2017.8019434

DO - 10.1109/ICME.2017.8019434

M3 - Conference contribution

AN - SCOPUS:85030229368

SP - 919

EP - 924

BT - 2017 IEEE International Conference on Multimedia and Expo, ICME 2017

T2 - 2017 IEEE International Conference on Multimedia and Expo, ICME 2017

Y2 - 10 July 2017 through 14 July 2017

ER -