Number of the records: 1  

Classification of Czech sign language alphabet letters using cnn – preliminary study

  1. 1.
    SYSNO ASEP0535189
    Document TypeC - Proceedings Paper (int. conf.)
    R&D Document TypeConference Paper
    TitleClassification of Czech sign language alphabet letters using cnn – preliminary study
    Author(s) Krejsa, Jiří (UT-L) RID, ORCID
    Věchet, Stanislav (UT-L) RID, ORCID
    Number of authors2
    Source TitleENGINEERING MECHANICS 2020. - Brno : Brno University of Technology Institute of Solid Mechanics, Mechatronics and Biomechanics, 2020 / Fuis V. - ISSN 1805-8248 - ISBN 978-80-214-5896-3
    Pagess. 310-313
    Number of pages4 s.
    Publication formPrint - P
    ActionInternational Conference Engineering Mechanics 2020 /26./
    Event date24.11.2020 - 25.11.2020
    VEvent locationBrno
    CountryCZ - Czech Republic
    Event typeWRD
    Languageeng - English
    CountryCZ - Czech Republic
    Keywordsclassification ; sign language ; convolution neural network
    Subject RIVIN - Informatics, Computer Science
    OECD categoryRobotics and automatic control
    Institutional supportUT-L - RVO:61388998
    UT WOS000667956100070
    AnnotationAbstract: The paper deals with the classification of Czech sign language single hand alphabet letters from static images using convolutional neural network (CNN). Proposed CNN architecture exhibits about 71% successful rate of classifying the letters signed by the person not included in the training data set.
    WorkplaceInstitute of Thermomechanics
    ContactMarie Kajprová,, Tel.: 266 053 154 ; Jana Lahovská,, Tel.: 266 053 823
    Year of Publishing2021
Number of the records: 1