AI at the Edge for Sign Language Learning Support

In the field of multimodal communication, sign language isand continues to be, one of the most understudied areas. Thanks to therecent advances in the field of deep learning, there are far-reaching implications and applications that neural networks can have for sign languagemastering. This paper describes a method for ASL alphabet recognitionusing Convolutional Neural Networks (CNN), which allows to monitoruser’s learning progress. American Sign Language (ASL) alphabet recognition by computer vision is a challenging task due to the complexity inASL signs, high interclass similarities, large intraclass variations, andconstant occlusions. We produced a robust model that classifies letterscorrectly in a majority of cases. The experimental results encouraged usto investigate the adoption of AI techniques to support learning of a signlanguage, as a natural language with its own syntax and lexicon. Thechallenge was to deliver a mobile sign language training solution thatusers may adopt during their everyday life. To satisfy the indispensableadditional computational resources to the locally connected end- userdevices, we propose the adoption of a Fog-Computing Architecture.