One Model is Not Enough: Ensembles for Isolated Sign Language Recognition

Show simple item record
dc.contributor.authorHrúz, Marek
dc.contributor.authorGruber, Ivan
dc.contributor.authorKanis, Jakub
dc.contributor.authorBoháček, Matyáš
dc.contributor.authorHlaváč, Miroslav
dc.contributor.authorKrňoul, Zdeněk
dc.date.accessioned2023-03-06T11:00:26Z
dc.date.available2023-03-06T11:00:26Z
dc.date.issued2022
dc.description.abstract-translatedIn this paper, we dive into sign language recognition, focusing on the recognition of isolated signs. The task is defined as a classification problem, where a sequence of frames (i.e., images) is recognized as one of the given sign language glosses. We analyze two appearance-based approaches, I3D and TimeSformer, and one pose-based approach, SPOTER. The appearance-based approaches are trained on a few different data modalities, whereas the performance of SPOTER is evaluated on different types of preprocessing. All the methods are tested on two publicly available datasets: AUTSL and WLASL300. We experiment with ensemble techniques to achieve new state-of-the-art results of 73.84% accuracy on the WLASL300 dataset by using the CMA-ES optimization method to find the best ensemble weight parameters. Furthermore, we present an ensembling technique based on the Transformer model, which we call Neural Ensembler.en
dc.format17 s.cs
dc.format.mimetypeapplication/pdf
dc.identifier.citationHRÚZ, M. GRUBER, I. KANIS, J. BOHÁČEK, M. HLAVÁČ, M. KRŇOUL, Z. One Model is Not Enough: Ensembles for Isolated Sign Language Recognition. SENSORS, 2022, roč. 22, č. 13, s. nestránkováno. ISSN: 1424-8220cs
dc.identifier.document-number824167200001
dc.identifier.doi10.3390/s22135043
dc.identifier.issn1424-8220
dc.identifier.obd43937108
dc.identifier.uri2-s2.0-85133217387
dc.identifier.urihttp://hdl.handle.net/11025/51652
dc.language.isoenen
dc.project.IDTN01000024/Národní centrum kompetence - Kybernetika a umělá inteligencecs
dc.project.ID90042/Velká výzkumná infrastruktura povinnost (J) - CESNET IIcs
dc.project.IDEF15_003/0000466/Umělá inteligence a uvažovánícs
dc.publisherMDPIen
dc.relation.ispartofseriesSENSORSen
dc.rights© authorsen
dc.rights.accessopenAccessen
dc.subject.translatedsign language recognitionen
dc.subject.translatedCNNen
dc.subject.translatedTransformeren
dc.subject.translatedensembleen
dc.titleOne Model is Not Enough: Ensembles for Isolated Sign Language Recognitionen
dc.typečlánekcs
dc.typearticleen
dc.type.statusPeer-revieweden
dc.type.versionpublishedVersionen

Files

Original bundle
Showing 1 - 1 out of 1 results
No Thumbnail Available
Name:
sensors-22-05043-v3-2.pdf
Size:
1.09 MB
Format:
Adobe Portable Document Format
Download

Collections

OBD
 Articles (KKY)
 Articles (NTIS)