## Image Novelty Detection Based on Mean-Shift and Typical Set Size

No Thumbnail Available
##### Files
There are no files associated with this item.
2022
##### Publication type
Contribution to a conference collection
Published
##### Published in
Image Analysis and Processing - ICIAP 2022 : 21st International Conference, Lecce, Italy, May 23-27, 2022, Proceedings, Part II / Sclaroff, Stan et al. (ed.). - Cham : Springer, 2022. - (Lecture Notes in Computer Science ; 13232). - pp. 755-766. - ISSN 0302-9743. - eISSN 1611-3349. - ISBN 978-3-031-06429-6
##### Abstract
The detection of anomalous or novel images given a training dataset of only clean reference data (inliers) is an important task in computer vision. We propose a new shallow approach that represents both inlier and outlier images as ensembles of patches, which allows us to effectively detect novelties as mean shifts between reference data and outliers with the Hotelling T2 test. Since mean-shift can only be detected when the outlier ensemble is sufficiently separate from the typical set of the inlier distribution, this typical set acts as a blind spot for novelty detection. We therefore minimize its estimated size as our selection rule for critical hyperparameters, such as, e.g., the size of the patches is crucial. To showcase the capabilities of our approach, we compare results with classical and deep learning methods on the popular datasets MNIST and CIFAR-10, and demonstrate its real-world applicability in a large-scale industrial inspection scenario.
##### Subject (DDC)
004 Computer Science
##### Conference
The International Conference on Image Analysis and Processing, ICIAP 2022, May 23, 2022 - May 27, 2022, Lecce, Italy
##### Cite This
ISO 690HERMANN, Matthias, Bastian GOLDLÜCKE, Matthias O. FRANZ, 2022. Image Novelty Detection Based on Mean-Shift and Typical Set Size. The International Conference on Image Analysis and Processing, ICIAP 2022. Lecce, Italy, May 23, 2022 - May 27, 2022. In: SCLAROFF, Stan, ed. and others. Image Analysis and Processing - ICIAP 2022 : 21st International Conference, Lecce, Italy, May 23-27, 2022, Proceedings, Part II. Cham:Springer, pp. 755-766. ISSN 0302-9743. eISSN 1611-3349. ISBN 978-3-031-06429-6. Available under: doi: 10.1007/978-3-031-06430-2_63
BibTex
@inproceedings{Hermann2022Image-59120,
year={2022},
doi={10.1007/978-3-031-06430-2_63},
title={Image Novelty Detection Based on Mean-Shift and Typical Set Size},
number={13232},
isbn={978-3-031-06429-6},
issn={0302-9743},
publisher={Springer},
series={Lecture Notes in Computer Science},
booktitle={Image Analysis and Processing - ICIAP 2022 : 21st International Conference, Lecce, Italy, May 23-27, 2022, Proceedings, Part II},
pages={755--766},
editor={Sclaroff, Stan},
author={Hermann, Matthias and Goldlücke, Bastian and Franz, Matthias O.}
}

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#" >
<dcterms:issued>2022</dcterms:issued>
<dcterms:available rdf:datatype="http://www.w3.org/2001/XMLSchema#dateTime">2022-11-11T09:39:03Z</dcterms:available>
<foaf:homepage rdf:resource="http://localhost:8080/"/>
<dc:creator>Franz, Matthias O.</dc:creator>
<dc:contributor>Franz, Matthias O.</dc:contributor>
<dcterms:abstract xml:lang="eng">The detection of anomalous or novel images given a training dataset of only clean reference data (inliers) is an important task in computer vision. We propose a new shallow approach that represents both inlier and outlier images as ensembles of patches, which allows us to effectively detect novelties as mean shifts between reference data and outliers with the Hotelling T2 test. Since mean-shift can only be detected when the outlier ensemble is sufficiently separate from the typical set of the inlier distribution, this typical set acts as a blind spot for novelty detection. We therefore minimize its estimated size as our selection rule for critical hyperparameters, such as, e.g., the size of the patches is crucial. To showcase the capabilities of our approach, we compare results with classical and deep learning methods on the popular datasets MNIST and CIFAR-10, and demonstrate its real-world applicability in a large-scale industrial inspection scenario.</dcterms:abstract>
<dc:contributor>Goldlücke, Bastian</dc:contributor>
<dcterms:title>Image Novelty Detection Based on Mean-Shift and Typical Set Size</dcterms:title>
<dc:date rdf:datatype="http://www.w3.org/2001/XMLSchema#dateTime">2022-11-11T09:39:03Z</dc:date>
<dc:creator>Hermann, Matthias</dc:creator>
<void:sparqlEndpoint rdf:resource="http://localhost/fuseki/dspace/sparql"/>
<dc:creator>Goldlücke, Bastian</dc:creator>
<dc:language>eng</dc:language>
<bibo:uri rdf:resource="https://kops.uni-konstanz.de/handle/123456789/59120"/>
<dcterms:isPartOf rdf:resource="https://kops.uni-konstanz.de/server/rdf/resource/123456789/36"/>
<dc:contributor>Hermann, Matthias</dc:contributor>
<dspace:isPartOfCollection rdf:resource="https://kops.uni-konstanz.de/server/rdf/resource/123456789/36"/>
</rdf:Description>
</rdf:RDF>

Yes