The study focuses on the issue of automatic sign language recognition, specifically Argentinian Sign Language (LSA), using a technique called ProbSom. The complexity of this issue lies in its multidisciplinary nature, involving areas such as video processing, image processing, intelligent systems and linguistics. The researchers developed a database of handshapes for LSA and used ProbSom, a supervised adaptation of self-organizing maps, for image processing and handshape classification. The technique achieved an accuracy rate above 90%. This research contributes to the broader goal of creating robust sign language recognition systems to aid the integration of hearing-impaired individuals.
Publication date: 27 Oct 2023
Project Page: http://www.lidi.info.unlp.edu.ar/
Paper: https://arxiv.org/pdf/2310.17427
