Redes Neurais Artificiais Aplicadas ao Reconhecimento de Expressões Faciais Negativas na Língua Brasileira de Sinais

Abstract

Facial expressions play an important role in Brazilian sign language, a gesture-visual language used by people with hearing disabilities. Aiming at collaborating with solutions to the automatic recognition of this language, facial expressions must also be considered. In this work, several artificial neural networks were trained and tested for negative facial expressions classification, whose detection in previous works had low performance. The results obtained indicate a 63% percentual improvement in this task when compared to existing works, with advances also in detection of conditional and binary facial expressions.

Publication
In: IV Escola Regional de Informática Norte I. Manaus, Amazonas.
Date