Publication detail
Czech Sign Language Single Hand Alphabet Classification with MediaPipe
ŠNAJDER, J. BEDNAŘÍK, J.
English title
Czech Sign Language Single Hand Alphabet Classification with MediaPipe
Type
article in a collection out of WoS and Scopus
Language
en
Original abstract
The paper presents the classification of static images of the single-handed Czech sign language alphabet. It uses the framework MediaPipe for annotation, and the classification is performed by a neural network using the TensorFlow computational library. The flow of the proposed method, data acquisition, preprocessing, and training are described in the paper. Obtained results consist of the classification success rate of the validation dataset for various MediaPipe configurations. The overall success rate was around 94%.
English abstract
The paper presents the classification of static images of the single-handed Czech sign language alphabet. It uses the framework MediaPipe for annotation, and the classification is performed by a neural network using the TensorFlow computational library. The flow of the proposed method, data acquisition, preprocessing, and training are described in the paper. Obtained results consist of the classification success rate of the validation dataset for various MediaPipe configurations. The overall success rate was around 94%.
Keywords in English
Czech sign language, Fingerspelling, Classification, Mediapipe, Neural network
Released
09.05.2022
Publisher
Institute of Theoretical and Applied Mechanics of the Czech Academy of Sciences
Location
Prague
ISBN
ISBN 978-80-86246-51
ISSN
1805-8256
Book
ENGINEERING MECHANICS 2022
Edition number
1
Pages from–to
381–384
Pages count
4
BIBTEX
@inproceedings{BUT178377,
author="Jan {Šnajder} and Josef {Bednařík},
title="Czech Sign Language Single Hand Alphabet Classification with MediaPipe",
booktitle="ENGINEERING MECHANICS 2022",
year="2022",
month="May",
pages="381--384",
publisher="Institute of Theoretical and Applied Mechanics of the Czech Academy of Sciences",
address="Prague",
isbn="ISBN 978-80-86246-51",
issn="1805-8256"
}