Czech Sign Language Single Hand Alphabet Letters Classification

0536892 - ÚT 2021 RIV US eng C - Konferenční příspěvek (zahraniční konf.)
Krejsa, Jiří - Věchet, Stanislav
Czech Sign Language Single Hand Alphabet Letters Classification.
Proceedings of the 2020 19th International Conference on Mechatronics - Mechatronika, ME 2020. New York: Institute of Electrical and Electronics Engineers Inc., 2020 - (Maga, D.; Hajek, J.), s. 1-341. ISBN 978-172815601-9.
[International Conference on Mechatronics - Mechatronika /19./. Praha (CZ), 02.12.2020-04.12.2020]
Institucionální podpora: RVO:61388998
Klíčová slova: Czech sign language * convolution neural network * classification
Obor OECD: Automation and control systems
https://ieeexplore.ieee.org/document/9286667

The paper deals with the classification of images of Czech sign language alphabet, single handed version in particular, without diacritics. The classification is performed by convolution neural network using TensorFlow computational library. Network topology, data acquisition and automatic labelling and obtained results are described in the paper. The accuracy on the test data - captured images of a person not previously seen by the network – was over 87%.
Trvalý link: http://hdl.handle.net/11104/0316609