End Date: 01/01/2023
SL-ReDu focuses on developing innovative algorithms of sign video recognition in order to considerably advance the current state-of-the-art in the field, so that SL recognition technologies can be incorporated into systems that allow robust HCI via SL. This aim lies at the forefront of research in the areas of computer vision and machine learning, where recent deep learning based breakthroughs have not yet fully propagated to SL recognition. The project will explore such innovative algorithms primarily for the recognition of the Greek Sign Language (GSL) based on 2- and 3-D video data exploiting a suitable annotated dataset and language model. SL-ReDu will integrate the developed SL recognition technology into a prototype demonstrator system, focusing on a use case in education via evaluating student SL performance at the Department of Special Education of the University of Thessaly in the context of learning and testing for the compulsory course “Introduction to Greek Sign Language” of the department curriculum. The demonstrator will incorporate an end-user interaction environment through GSL, addressing the current lack of assessment tools for evaluating the level of SL competence of non-native users (students and learners), and more particularly the need to ensure testing credibility and consistency.