Improving and Extending Continuous Sign Language Recognition: Taking Iconicity and Spatial Language into account

Autor/a: BELISSEN, Valentin; GOUIFFES, Michèle; BRAFFORT, Annelies
Año: 2020
Editorial: ELRA
Tipo de código: Copyright
Soporte: Digital


Lingüística » Lingüística de otras Lenguas de Signos


In a series of recent papers focused on Continuous Sign Language Recognition (CSLR), the performance seems to be getting better and better and one may wonder what remains to be done before Sign Language Translation (SLT) can be envisioned. However, we want to show that the current trend in CSLR has inherent limitations, which prevent it from extending to SL understanding, a fortiori to SLT. Besides, extending on previous work, we try to highlight the value of recognizing illustrative elements in SL for the task of CSLR, even though few corpora are available for this research. This paper will also discuss the related challenges and the difficulties in evaluating the recognition results.

En Proceedings of the 9th Workshop on the Representation and Processing of Sign Languages, pp. 7-12.