Effective Aesthetics Prediction with Multi-level Spatially Pooled Features

Loading...
Thumbnail Image
Date
2019
Editors
Contact
Journal ISSN
Electronic ISSN
ISBN
Bibliographical data
Publisher
Series
DOI (citable link)
ArXiv-ID
International patent number
Link to the license
EU project number
Project
Open Access publication
Restricted until
Title in another language
Research Projects
Organizational Units
Journal Issue
Publication type
Preprint
Publication status
Published
Published in
Abstract
We propose an effective deep learning approach to aesthetics quality assessment that relies on a new type of pre-trained features, and apply it to the AVA data set, the currently largest aesthetics database. While previous approaches miss some of the information in the original images, due to taking small crops, down-scaling or warping the originals during training, we propose the first method that efficiently supports full resolution images as an input, and can be trained on variable input sizes. This allows us to significantly improve upon the state of the art, increasing the Spearman rank-order correlation coefficient (SRCC) of ground-truth mean opinion scores (MOS) from the existing best reported of 0.612 to 0.756. To achieve this performance, we extract multi-level spatially pooled (MLSP) features from all convolutional blocks of a pre-trained InceptionResNet-v2 network, and train a custom shallow Convolutional Neural Network (CNN) architecture on these new features.
Summary in another language
Subject (DDC)
004 Computer Science
Keywords
Conference
Review
undefined / . - undefined, undefined. - (undefined; undefined)
Cite This
ISO 690HOSU, Vlad, Bastian GOLDLÜCKE, Dietmar SAUPE, 2019. Effective Aesthetics Prediction with Multi-level Spatially Pooled Features
BibTex
@unpublished{Hosu2019-04-02T12:58:12ZEffec-45609,
  year={2019},
  title={Effective Aesthetics Prediction with Multi-level Spatially Pooled Features},
  author={Hosu, Vlad and Goldlücke, Bastian and Saupe, Dietmar},
  note={To appear in CVPR 2019}
}
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/123456789/45609">
    <dc:contributor>Saupe, Dietmar</dc:contributor>
    <dcterms:issued>2019-04-02T12:58:12Z</dcterms:issued>
    <dc:creator>Goldlücke, Bastian</dc:creator>
    <foaf:homepage rdf:resource="http://localhost:8080/"/>
    <dcterms:available rdf:datatype="http://www.w3.org/2001/XMLSchema#dateTime">2019-04-09T13:48:49Z</dcterms:available>
    <dcterms:rights rdf:resource="https://rightsstatements.org/page/InC/1.0/"/>
    <dspace:isPartOfCollection rdf:resource="https://kops.uni-konstanz.de/server/rdf/resource/123456789/36"/>
    <dc:language>eng</dc:language>
    <void:sparqlEndpoint rdf:resource="http://localhost/fuseki/dspace/sparql"/>
    <bibo:uri rdf:resource="https://kops.uni-konstanz.de/handle/123456789/45609"/>
    <dcterms:hasPart rdf:resource="https://kops.uni-konstanz.de/bitstream/123456789/45609/1/Hosu_2-amv1uzjjhj4x0.pdf"/>
    <dc:rights>terms-of-use</dc:rights>
    <dc:creator>Saupe, Dietmar</dc:creator>
    <dcterms:title>Effective Aesthetics Prediction with Multi-level Spatially Pooled Features</dcterms:title>
    <dcterms:isPartOf rdf:resource="https://kops.uni-konstanz.de/server/rdf/resource/123456789/36"/>
    <dc:date rdf:datatype="http://www.w3.org/2001/XMLSchema#dateTime">2019-04-09T13:48:49Z</dc:date>
    <dc:contributor>Hosu, Vlad</dc:contributor>
    <dcterms:abstract xml:lang="eng">We propose an effective deep learning approach to aesthetics quality assessment that relies on a new type of pre-trained features, and apply it to the AVA data set, the currently largest aesthetics database. While previous approaches miss some of the information in the original images, due to taking small crops, down-scaling or warping the originals during training, we propose the first method that efficiently supports full resolution images as an input, and can be trained on variable input sizes. This allows us to significantly improve upon the state of the art, increasing the Spearman rank-order correlation coefficient (SRCC) of ground-truth mean opinion scores (MOS) from the existing best reported of 0.612 to 0.756. To achieve this performance, we extract multi-level spatially pooled (MLSP) features from all convolutional blocks of a pre-trained InceptionResNet-v2 network, and train a custom shallow Convolutional Neural Network (CNN) architecture on these new features.</dcterms:abstract>
    <dc:creator>Hosu, Vlad</dc:creator>
    <dspace:hasBitstream rdf:resource="https://kops.uni-konstanz.de/bitstream/123456789/45609/1/Hosu_2-amv1uzjjhj4x0.pdf"/>
    <dc:contributor>Goldlücke, Bastian</dc:contributor>
  </rdf:Description>
</rdf:RDF>
Internal note
xmlui.Submission.submit.DescribeStep.inputForms.label.kops_note_fromSubmitter
Contact
URL of original publication
Test date of URL
Examination date of dissertation
Method of financing
Comment on publication
To appear in CVPR 2019
Alliance license
Corresponding Authors der Uni Konstanz vorhanden
International Co-Authors
Bibliography of Konstanz
Yes
Refereed