Deep Hough-Transform Line Priors

Y. Lin; S. Pintea; J.C. van Gemert

doi:10.1007/978-3-030-58542-6_20

Deep Hough-Transform Line Priors

Pattern Recognition and Bioinformatics

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

26 Citations (Scopus)

Abstract

Classical work on line segment detection is knowledge-based; it uses carefully designed geometric priors using either image gradients, pixel groupings, or Hough transform variants. Instead, current deep learning methods do away with all prior knowledge and replace priors by training deep networks on large manually annotated datasets. Here, we reduce the dependency on labeled data by building on the classic knowledge-based priors while using deep networks to learn features. We add line priors through a trainable Hough transform block into a deep network. Hough transform provides the prior knowledge about global line parameterizations, while the convolutional layers can learn the local gradient-like line features. On the Wireframe (ShanghaiTech) and York Urban datasets we show that adding prior knowledge improves data efficiency as line priors no longer need to be learned from data.

Original language	English
Title of host publication	Computer Vision – ECCV 2020 - 16th European Conference, 2020, Proceedings
Editors	Andrea Vedaldi, Horst Bischof, Thomas Brox, Jan-Michael Frahm
Place of Publication	European Conference on Computer Vision ECCV 2020: Computer Vision – ECCV 2020
Publisher	Springer
Pages	323-340
Number of pages	18
Volume	12367
ISBN (Electronic)	978-3-030-58542-6
ISBN (Print)	978-3-030-58541-9
DOIs	https://doi.org/10.1007/978-3-030-58542-6_20
Publication status	Published - 2020

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	12367 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Bibliographical note

@article{lin2020deep,
title={Deep Hough-Transform Line Priors},
author={Lin, Yancong and Pintea, Silvia L and van Gemert, Jan C},
booktitle={EECV 2020},
year={2020}
}

Keywords

Global line prior
Hough transform
Line segment detection

Access to Document

10.1007/978-3-030-58542-6_20

Cite this

Lin, Y., Pintea, S., & van Gemert, J. C. (2020). Deep Hough-Transform Line Priors. In A. Vedaldi, H. Bischof, T. Brox, & J.-M. Frahm (Eds.), Computer Vision – ECCV 2020 - 16th European Conference, 2020, Proceedings (Vol. 12367, pp. 323-340). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12367 LNCS). Springer. https://doi.org/10.1007/978-3-030-58542-6_20

Lin, Y. ; Pintea, S. ; van Gemert, J.C. / Deep Hough-Transform Line Priors. Computer Vision – ECCV 2020 - 16th European Conference, 2020, Proceedings. editor / Andrea Vedaldi ; Horst Bischof ; Thomas Brox ; Jan-Michael Frahm. Vol. 12367 European Conference on Computer Vision ECCV 2020: Computer Vision – ECCV 2020 : Springer, 2020. pp. 323-340 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{4a6bdcec721c4aeba0c82a76a9038813,

title = "Deep Hough-Transform Line Priors",

abstract = "Classical work on line segment detection is knowledge-based; it uses carefully designed geometric priors using either image gradients, pixel groupings, or Hough transform variants. Instead, current deep learning methods do away with all prior knowledge and replace priors by training deep networks on large manually annotated datasets. Here, we reduce the dependency on labeled data by building on the classic knowledge-based priors while using deep networks to learn features. We add line priors through a trainable Hough transform block into a deep network. Hough transform provides the prior knowledge about global line parameterizations, while the convolutional layers can learn the local gradient-like line features. On the Wireframe (ShanghaiTech) and York Urban datasets we show that adding prior knowledge improves data efficiency as line priors no longer need to be learned from data.",

keywords = "Global line prior, Hough transform, Line segment detection",

author = "Y. Lin and S. Pintea and {van Gemert}, J.C.",

note = "@article{lin2020deep, title={Deep Hough-Transform Line Priors}, author={Lin, Yancong and Pintea, Silvia L and van Gemert, Jan C}, booktitle={EECV 2020}, year={2020} }",

year = "2020",

doi = "10.1007/978-3-030-58542-6_20",

language = "English",

isbn = "978-3-030-58541-9",

volume = "12367",

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

publisher = "Springer",

pages = "323--340",

editor = "Andrea Vedaldi and Horst Bischof and Thomas Brox and Jan-Michael Frahm",

booktitle = "Computer Vision – ECCV 2020 - 16th European Conference, 2020, Proceedings",

}

Lin, Y , Pintea, S & van Gemert, JC 2020, Deep Hough-Transform Line Priors. in A Vedaldi, H Bischof, T Brox & J-M Frahm (eds), Computer Vision – ECCV 2020 - 16th European Conference, 2020, Proceedings. vol. 12367, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12367 LNCS, Springer, European Conference on Computer Vision ECCV 2020: Computer Vision – ECCV 2020, pp. 323-340. https://doi.org/10.1007/978-3-030-58542-6_20

Deep Hough-Transform Line Priors. / Lin, Y.; Pintea, S.; van Gemert, J.C.
Computer Vision – ECCV 2020 - 16th European Conference, 2020, Proceedings. ed. / Andrea Vedaldi; Horst Bischof; Thomas Brox; Jan-Michael Frahm. Vol. 12367 European Conference on Computer Vision ECCV 2020: Computer Vision – ECCV 2020: Springer, 2020. p. 323-340 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12367 LNCS).

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Deep Hough-Transform Line Priors

AU - Lin, Y.

AU - Pintea, S.

AU - van Gemert, J.C.

N1 - @article{lin2020deep, title={Deep Hough-Transform Line Priors}, author={Lin, Yancong and Pintea, Silvia L and van Gemert, Jan C}, booktitle={EECV 2020}, year={2020} }

PY - 2020

Y1 - 2020

N2 - Classical work on line segment detection is knowledge-based; it uses carefully designed geometric priors using either image gradients, pixel groupings, or Hough transform variants. Instead, current deep learning methods do away with all prior knowledge and replace priors by training deep networks on large manually annotated datasets. Here, we reduce the dependency on labeled data by building on the classic knowledge-based priors while using deep networks to learn features. We add line priors through a trainable Hough transform block into a deep network. Hough transform provides the prior knowledge about global line parameterizations, while the convolutional layers can learn the local gradient-like line features. On the Wireframe (ShanghaiTech) and York Urban datasets we show that adding prior knowledge improves data efficiency as line priors no longer need to be learned from data.

AB - Classical work on line segment detection is knowledge-based; it uses carefully designed geometric priors using either image gradients, pixel groupings, or Hough transform variants. Instead, current deep learning methods do away with all prior knowledge and replace priors by training deep networks on large manually annotated datasets. Here, we reduce the dependency on labeled data by building on the classic knowledge-based priors while using deep networks to learn features. We add line priors through a trainable Hough transform block into a deep network. Hough transform provides the prior knowledge about global line parameterizations, while the convolutional layers can learn the local gradient-like line features. On the Wireframe (ShanghaiTech) and York Urban datasets we show that adding prior knowledge improves data efficiency as line priors no longer need to be learned from data.

KW - Global line prior

KW - Hough transform

KW - Line segment detection

UR - http://www.scopus.com/inward/record.url?scp=85097257100&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-58542-6_20

DO - 10.1007/978-3-030-58542-6_20

M3 - Conference contribution

SN - 978-3-030-58541-9

VL - 12367

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

SP - 323

EP - 340

BT - Computer Vision – ECCV 2020 - 16th European Conference, 2020, Proceedings

A2 - Vedaldi, Andrea

A2 - Bischof, Horst

A2 - Brox, Thomas

A2 - Frahm, Jan-Michael

PB - Springer

CY - European Conference on Computer Vision ECCV 2020: Computer Vision – ECCV 2020

ER -

Lin Y , Pintea S , van Gemert JC. Deep Hough-Transform Line Priors. In Vedaldi A, Bischof H, Brox T, Frahm JM, editors, Computer Vision – ECCV 2020 - 16th European Conference, 2020, Proceedings. Vol. 12367. European Conference on Computer Vision ECCV 2020: Computer Vision – ECCV 2020: Springer. 2020. p. 323-340. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-58542-6_20

Deep Hough-Transform Line Priors

Abstract

Publication series

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this