Publikation:

Tarsier : Evolving Noise Injection in Super-Resolution GANs

Lade...
Vorschaubild

Dateien

Zu diesem Dokument gibt es keine Dateien.

Datum

2021

Autor:innen

Roziere, Baptiste
Carraz Rakotonirina, Nathanael
Rasoanaivo, Andry
Couprie, Camille
Teytaud, Olivier

Herausgeber:innen

Kontakt

ISSN der Zeitschrift

Electronic ISSN

ISBN

Bibliografische Daten

Verlag

Schriftenreihe

Auflagebezeichnung

URI (zitierfähiger Link)
ArXiv-ID

Internationale Patentnummer

Angaben zur Forschungsförderung

Projekt

Open Access-Veröffentlichung
Core Facility der Universität Konstanz

Gesperrt bis

Titel in einer weiteren Sprache

Publikationstyp
Beitrag zu einem Konferenzband
Publikationsstatus
Published

Erschienen in

Proceedings of ICPR 2020 : 25th International Conference on Pattern Recognition. Piscataway, NJ: IEEE, 2021, pp. 7028-7035. ISBN 978-1-72818-808-9. Available under: doi: 10.1109/ICPR48806.2021.9413318

Zusammenfassung

Super-resolution aims at increasing the resolution and level of detail within an image. The current state of the art in general single-image super-resolution is held by NESRGAN+, which injects a Gaussian noise after each residual layer at training time. In this paper, we harness evolutionary methods to improve NESRGAN+ by optimizing the noise injection at inference time. More precisely, we use Diagonal CMA to optimize the injected noise according to a novel criterion combining quality assessment and realism. Our results are validated by the PIRM perceptual score and a human study. Our method outperforms NESRGAN+ on several standard super-resolution datasets. More generally, our approach can be used to optimize any method based on noise injection.

Zusammenfassung in einer weiteren Sprache

Fachgebiet (DDC)
004 Informatik

Schlagwörter

Konferenz

ICPR 2020 : 25th International Conference on Pattern Recognition, 10. Jan. 2021 - 15. Jan. 2021, Milan, Italy
Rezension
undefined / . - undefined, undefined

Forschungsvorhaben

Organisationseinheiten

Zeitschriftenheft

Zugehörige Datensätze in KOPS

Zitieren

ISO 690ROZIERE, Baptiste, Nathanael CARRAZ RAKOTONIRINA, Vlad HOSU, Andry RASOANAIVO, Hanhe LIN, Camille COUPRIE, Olivier TEYTAUD, 2021. Tarsier : Evolving Noise Injection in Super-Resolution GANs. ICPR 2020 : 25th International Conference on Pattern Recognition. Milan, Italy, 10. Jan. 2021 - 15. Jan. 2021. In: Proceedings of ICPR 2020 : 25th International Conference on Pattern Recognition. Piscataway, NJ: IEEE, 2021, pp. 7028-7035. ISBN 978-1-72818-808-9. Available under: doi: 10.1109/ICPR48806.2021.9413318
BibTex
@inproceedings{Roziere2021Tarsi-55115,
  year={2021},
  doi={10.1109/ICPR48806.2021.9413318},
  title={Tarsier : Evolving Noise Injection in Super-Resolution GANs},
  isbn={978-1-72818-808-9},
  publisher={IEEE},
  address={Piscataway, NJ},
  booktitle={Proceedings of ICPR 2020 : 25th International Conference on Pattern Recognition},
  pages={7028--7035},
  author={Roziere, Baptiste and Carraz Rakotonirina, Nathanael and Hosu, Vlad and Rasoanaivo, Andry and Lin, Hanhe and Couprie, Camille and Teytaud, Olivier}
}
RDF
<rdf:RDF
    xmlns:dcterms="http://purl.org/dc/terms/"
    xmlns:dc="http://purl.org/dc/elements/1.1/"
    xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
    xmlns:bibo="http://purl.org/ontology/bibo/"
    xmlns:dspace="http://digital-repositories.org/ontologies/dspace/0.1.0#"
    xmlns:foaf="http://xmlns.com/foaf/0.1/"
    xmlns:void="http://rdfs.org/ns/void#"
    xmlns:xsd="http://www.w3.org/2001/XMLSchema#" > 
  <rdf:Description rdf:about="https://kops.uni-konstanz.de/server/rdf/resource/123456789/55115">
    <dc:creator>Teytaud, Olivier</dc:creator>
    <dc:creator>Lin, Hanhe</dc:creator>
    <dc:creator>Rasoanaivo, Andry</dc:creator>
    <dc:contributor>Lin, Hanhe</dc:contributor>
    <dspace:isPartOfCollection rdf:resource="https://kops.uni-konstanz.de/server/rdf/resource/123456789/36"/>
    <foaf:homepage rdf:resource="http://localhost:8080/"/>
    <dc:contributor>Hosu, Vlad</dc:contributor>
    <dcterms:available rdf:datatype="http://www.w3.org/2001/XMLSchema#dateTime">2021-10-01T10:59:49Z</dcterms:available>
    <dcterms:abstract xml:lang="eng">Super-resolution aims at increasing the resolution and level of detail within an image. The current state of the art in general single-image super-resolution is held by NESRGAN+, which injects a Gaussian noise after each residual layer at training time. In this paper, we harness evolutionary methods to improve NESRGAN+ by optimizing the noise injection at inference time. More precisely, we use Diagonal CMA to optimize the injected noise according to a novel criterion combining quality assessment and realism. Our results are validated by the PIRM perceptual score and a human study. Our method outperforms NESRGAN+ on several standard super-resolution datasets. More generally, our approach can be used to optimize any method based on noise injection.</dcterms:abstract>
    <dc:creator>Carraz Rakotonirina, Nathanael</dc:creator>
    <bibo:uri rdf:resource="https://kops.uni-konstanz.de/handle/123456789/55115"/>
    <dc:contributor>Roziere, Baptiste</dc:contributor>
    <dc:contributor>Couprie, Camille</dc:contributor>
    <dc:date rdf:datatype="http://www.w3.org/2001/XMLSchema#dateTime">2021-10-01T10:59:49Z</dc:date>
    <dc:creator>Roziere, Baptiste</dc:creator>
    <dc:contributor>Teytaud, Olivier</dc:contributor>
    <dcterms:issued>2021</dcterms:issued>
    <dcterms:title>Tarsier : Evolving Noise Injection in Super-Resolution GANs</dcterms:title>
    <dcterms:isPartOf rdf:resource="https://kops.uni-konstanz.de/server/rdf/resource/123456789/36"/>
    <dc:contributor>Carraz Rakotonirina, Nathanael</dc:contributor>
    <void:sparqlEndpoint rdf:resource="http://localhost/fuseki/dspace/sparql"/>
    <dc:creator>Hosu, Vlad</dc:creator>
    <dc:language>eng</dc:language>
    <dc:contributor>Rasoanaivo, Andry</dc:contributor>
    <dc:creator>Couprie, Camille</dc:creator>
  </rdf:Description>
</rdf:RDF>

Interner Vermerk

xmlui.Submission.submit.DescribeStep.inputForms.label.kops_note_fromSubmitter

Kontakt
URL der Originalveröffentl.

Prüfdatum der URL

Prüfungsdatum der Dissertation

Finanzierungsart

Kommentar zur Publikation

Allianzlizenz
Corresponding Authors der Uni Konstanz vorhanden
Internationale Co-Autor:innen
Universitätsbibliographie
Ja
Begutachtet
Diese Publikation teilen