Show simple item record

dc.contributor.author
Mueller, Markus S.
dc.contributor.author
Sattler, Torsten
dc.contributor.author
Pollefeys, Marc
dc.contributor.author
Jutzi, Boris
dc.contributor.editor
Stilla, Uwe
dc.contributor.editor
Hoegner, Ludwig
dc.contributor.editor
Xu, Yusheng
dc.date.accessioned
2020-01-17T11:56:16Z
dc.date.available
2020-01-16T13:40:38Z
dc.date.available
2020-01-17T11:46:01Z
dc.date.available
2020-01-17T11:56:16Z
dc.date.issued
2019
dc.identifier.issn
2194-9042
dc.identifier.issn
2194-9050
dc.identifier.other
10.5194/isprs-annals-iv-2-w7-111-2019
en_US
dc.identifier.uri
http://hdl.handle.net/20.500.11850/391178
dc.identifier.doi
10.3929/ethz-b-000391178
dc.description.abstract
The performance of machine learning and deep learning algorithms for image analysis depends significantly on the quantity and quality of the training data. The generation of annotated training data is often costly, time-consuming and laborious. Data augmentation is a powerful option to overcome these drawbacks. Therefore, we augment training data by rendering images with arbitrary poses from 3D models to increase the quantity of training images. These training images usually show artifacts and are of limited use for advanced image analysis. Therefore, we propose to use image-to-image translation to transform images from a rendered domain to a captured domain. We show that translated images in the captured domain are of higher quality than the rendered images. Moreover, we demonstrate that image-to-image translation based on rendered 3D models enhances the performance of common computer vision tasks, namely feature matching, image retrieval and visual localization. The experimental results clearly show the enhancement on translated images over rendered images for all investigated tasks. In addition to this, we present the advantages utilizing translated images over exclusively captured images for visual localization.
en_US
dc.format
application/pdf
en_US
dc.language.iso
en
en_US
dc.publisher
Copernicus
en_US
dc.rights.uri
http://creativecommons.org/licenses/by/4.0/
dc.subject
Image-to-Image Translation
en_US
dc.subject
Convolutional Neural Networks
en_US
dc.subject
Generative Adversarial Networks
en_US
dc.subject
Data Augmentation
en_US
dc.subject
3D Models
en_US
dc.subject
Feature Matching
en_US
dc.subject
Image Retrieval
en_US
dc.subject
Visual Localization
en_US
dc.title
Image-to-image translation for enhanced feature matching, image retrieval and visual localization
en_US
dc.type
Conference Paper
dc.rights.license
Creative Commons Attribution 4.0 International
dc.date.published
2019-09-16
ethz.journal.title
ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
ethz.journal.volume
IV-2/W7
en_US
ethz.journal.abbreviated
ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci.
ethz.pages.start
111
en_US
ethz.pages.end
119
en_US
ethz.version.deposit
publishedVersion
en_US
ethz.event
Photogrammetric Image Analysis & Munich Remote Sensing Symposium (PIA19+MRSS19)
en_US
ethz.event.location
Munich, Germany
en_US
ethz.event.date
September 18-20, 2019
en_US
ethz.identifier.scopus
ethz.publication.place
Göttingen
en_US
ethz.publication.status
published
en_US
ethz.leitzahl
ETH Zürich::00002 - ETH Zürich::00012 - Lehre und Forschung::00007 - Departemente::02150 - Dep. Informatik / Dep. of Computer Science::02659 - Institut für Visual Computing / Institute for Visual Computing::03766 - Pollefeys, Marc / Pollefeys, Marc
en_US
ethz.leitzahl.certified
ETH Zürich::00002 - ETH Zürich::00012 - Lehre und Forschung::00007 - Departemente::02150 - Dep. Informatik / Dep. of Computer Science::02659 - Institut für Visual Computing / Institute for Visual Computing::03766 - Pollefeys, Marc / Pollefeys, Marc
en_US
ethz.date.deposited
2020-01-16T13:40:45Z
ethz.source
FORM
ethz.eth
yes
en_US
ethz.availability
Open access
en_US
ethz.rosetta.installDate
2020-01-17T11:56:32Z
ethz.rosetta.lastUpdated
2021-02-15T07:22:51Z
ethz.rosetta.versionExported
true
ethz.COinS
ctx_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.atitle=Image-to-image%20translation%20for%20enhanced%20feature%20matching,%20image%20retrieval%20and%20visual%20localization&rft.jtitle=ISPRS%20Annals%20of%20the%20Photogrammetry,%20Remote%20Sensing%20and%20Spatial%20Information%20Sciences&rft.date=2019&rft.volume=IV-2/W7&rft.spage=111&rft.epage=119&rft.issn=2194-9042&2194-9050&rft.au=Mueller,%20Markus%20S.&Sattler,%20Torsten&Pollefeys,%20Marc&Jutzi,%20Boris&rft.genre=proceeding&rft_id=info:doi/10.5194/isprs-annals-iv-2-w7-111-2019&
 Search print copy at ETH Library

Files in this item

Thumbnail

Publication type

Show simple item record