This research focuses on enabling the development of (AI-based) technology that can translate from en to (Flemish) sign language. Our research and valorization goals are oriented towards applications that are supported and desired by the stakeholders (mainly the Flemish deaf and hard of hearing and their relatives) and that are technologically feasible on the short and medium-term. We focus on removing a number of communication thresholds in circumstances where a sign language interpreter is not immediately available. 

Our research is performed within a close collaboration with the Flemish deaf and hard of hearing community and experts in Flemish sign language linguistics. While our first societally relevant results are being valorized, our research will concentrate on the next challenges: accessibility tools and digital tools for sign language learners. 

In contrast to machine translation for written and spoken languages, the amount of high-quality labelled data available for sign language is extremely low. The main research questions in this project are therefore aimed at overcoming this limitation:

• Data-efficient techniques for sign language classification and translation
• Maximal dat areuse accross sign languages
• Identification and extraction of a minimal subset of relevant information from sign language videos