S2DNet: Learning Image Features for Accurate Sparse-to-Dense Matching

Hugo Germain; Guillaume Bourmaud; Vincent Lepetit

doi:10.1007/978-3-030-58580-8_37

S2DNet: Learning Image Features for Accurate Sparse-to-Dense Matching

Hugo Germain^*, Guillaume Bourmaud, Vincent Lepetit

^*Corresponding author for this work

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

Abstract

Establishing robust and accurate correspondences is a fundamental backbone to many computer vision algorithms. While recent learning-based feature matching methods have shown promising results in providing robust correspondences under challenging conditions, they are often limited in terms of precision. In this paper, we introduce S2DNet, a novel feature matching pipeline, designed and trained to efficiently establish both robust and accurate correspondences. By leveraging a sparse-to-dense matching paradigm, we cast the correspondence learning problem as a supervised classification task to learn to output highly peaked correspondence maps. We show that S2DNet achieves state-of-the-art results on the HPatches benchmark, as well as on several long-term visual localization datasets.

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
Publisher	Springer, Cham
Pages	626-643
Number of pages	18
ISBN (Print)	9783030585792
DOIs	https://doi.org/10.1007/978-3-030-58580-8_37
Publication status	Published - 23 Aug 2020
Externally published	Yes
Event	16th European Conference on Computer Vision: ECCV 2020 - Virtual, Glasgow, United Kingdom Duration: 23 Aug 2020 → 28 Aug 2020

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	12348 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	16th European Conference on Computer Vision
Abbreviated title	ECCV 2020
Country/Territory	United Kingdom
City	Virtual, Glasgow
Period	23/08/20 → 28/08/20

Keywords

Classification
Feature matching
Visual localization

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-030-58580-8_37

Cite this

Germain, H., Bourmaud, G., & Lepetit, V. (2020). S2DNet: Learning Image Features for Accurate Sparse-to-Dense Matching. In A. Vedaldi, H. Bischof, T. Brox, & J.-M. Frahm (Eds.), Computer Vision – ECCV 2020 - 16th European Conference 2020, Proceedings (pp. 626-643). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12348 LNCS). Springer, Cham. https://doi.org/10.1007/978-3-030-58580-8_37

S2DNet: Learning Image Features for Accurate Sparse-to-Dense Matching. / Germain, Hugo; Bourmaud, Guillaume; Lepetit, Vincent.
Computer Vision – ECCV 2020 - 16th European Conference 2020, Proceedings. ed. / Andrea Vedaldi; Horst Bischof; Thomas Brox; Jan-Michael Frahm. Springer, Cham, 2020. p. 626-643 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12348 LNCS).

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

Germain, H, Bourmaud, G & Lepetit, V 2020, S2DNet: Learning Image Features for Accurate Sparse-to-Dense Matching. in A Vedaldi, H Bischof, T Brox & J-M Frahm (eds), Computer Vision – ECCV 2020 - 16th European Conference 2020, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12348 LNCS, Springer, Cham, pp. 626-643, 16th European Conference on Computer Vision, Virtual, Glasgow, United Kingdom, 23/08/20. https://doi.org/10.1007/978-3-030-58580-8_37

Germain H, Bourmaud G, Lepetit V. S2DNet: Learning Image Features for Accurate Sparse-to-Dense Matching. In Vedaldi A, Bischof H, Brox T, Frahm JM, editors, Computer Vision – ECCV 2020 - 16th European Conference 2020, Proceedings. Springer, Cham. 2020. p. 626-643. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-58580-8_37

Germain, Hugo ; Bourmaud, Guillaume ; Lepetit, Vincent. / S2DNet: Learning Image Features for Accurate Sparse-to-Dense Matching. Computer Vision – ECCV 2020 - 16th European Conference 2020, Proceedings. editor / Andrea Vedaldi ; Horst Bischof ; Thomas Brox ; Jan-Michael Frahm. Springer, Cham, 2020. pp. 626-643 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{7fe0435dcd7d4b7c89874b1acf1f41a8,

title = "S2DNet: Learning Image Features for Accurate Sparse-to-Dense Matching",

abstract = "Establishing robust and accurate correspondences is a fundamental backbone to many computer vision algorithms. While recent learning-based feature matching methods have shown promising results in providing robust correspondences under challenging conditions, they are often limited in terms of precision. In this paper, we introduce S2DNet, a novel feature matching pipeline, designed and trained to efficiently establish both robust and accurate correspondences. By leveraging a sparse-to-dense matching paradigm, we cast the correspondence learning problem as a supervised classification task to learn to output highly peaked correspondence maps. We show that S2DNet achieves state-of-the-art results on the HPatches benchmark, as well as on several long-term visual localization datasets.",

keywords = "Classification, Feature matching, Visual localization",

author = "Hugo Germain and Guillaume Bourmaud and Vincent Lepetit",

year = "2020",

month = aug,

day = "23",

doi = "10.1007/978-3-030-58580-8_37",

language = "English",

isbn = "9783030585792",

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

publisher = "Springer, Cham",

pages = "626--643",

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

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

note = "16th European Conference on Computer Vision : ECCV 2020, ECCV 2020 ; Conference date: 23-08-2020 Through 28-08-2020",

}

TY - GEN

T1 - S2DNet: Learning Image Features for Accurate Sparse-to-Dense Matching

AU - Germain, Hugo

AU - Bourmaud, Guillaume

AU - Lepetit, Vincent

PY - 2020/8/23

Y1 - 2020/8/23

N2 - Establishing robust and accurate correspondences is a fundamental backbone to many computer vision algorithms. While recent learning-based feature matching methods have shown promising results in providing robust correspondences under challenging conditions, they are often limited in terms of precision. In this paper, we introduce S2DNet, a novel feature matching pipeline, designed and trained to efficiently establish both robust and accurate correspondences. By leveraging a sparse-to-dense matching paradigm, we cast the correspondence learning problem as a supervised classification task to learn to output highly peaked correspondence maps. We show that S2DNet achieves state-of-the-art results on the HPatches benchmark, as well as on several long-term visual localization datasets.

AB - Establishing robust and accurate correspondences is a fundamental backbone to many computer vision algorithms. While recent learning-based feature matching methods have shown promising results in providing robust correspondences under challenging conditions, they are often limited in terms of precision. In this paper, we introduce S2DNet, a novel feature matching pipeline, designed and trained to efficiently establish both robust and accurate correspondences. By leveraging a sparse-to-dense matching paradigm, we cast the correspondence learning problem as a supervised classification task to learn to output highly peaked correspondence maps. We show that S2DNet achieves state-of-the-art results on the HPatches benchmark, as well as on several long-term visual localization datasets.

KW - Classification

KW - Feature matching

KW - Visual localization

UR - http://www.scopus.com/inward/record.url?scp=85097842240&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-58580-8_37

DO - 10.1007/978-3-030-58580-8_37

M3 - Conference paper

SN - 9783030585792

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

SP - 626

EP - 643

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, Cham

T2 - 16th European Conference on Computer Vision

Y2 - 23 August 2020 through 28 August 2020

ER -

S2DNet: Learning Image Features for Accurate Sparse-to-Dense Matching

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Cite this