Eshara is Redefining Arabic Sign Language Translation with AI
Eshara, an AI-powered Arabic sign language translator, has won the 2024 UAE James Dyson Award. Developed by three engineering students from the American University of Sharjah (AUS) — Salma Shaarawi, Mohamed Eid, and Aley Eshra — Eshara addresses a critical issue in the Middle East: the lack of Arabic sign language translation services. This innovation aligns with the UAE’s vision of improving accessibility for people of determination, helping to bridge the communication gap for the hard of hearing in Arabic-speaking societies.
Unlike traditional translation services, which often lack support for Arabic sign language, Eshara is a pioneering platform tailored to the region’s needs. The system enhances interaction between hard-of-hearing individuals and the broader public, providing a solution where existing services fall short.
The James Dyson Award, which has challenged UAE students for the past seven years to "design something that solves a problem," encourages bold, innovative thinking. This year, the Eshara team recognised the communication barriers faced by hard-of-hearing individuals due to the absence of Arabic sign language resources. Inspired by a conversation with the African Sign Languages Resource Center (ASLRC) and driven by a vision of inclusivity, the team set out to change that.
Through extensive research and development, the students created a deep learning model by filming 62,000 videos of themselves signing. This led to the creation of a web application that translates sign language gestures into text in real time using a phone camera, making it widely accessible through standard web browsers. Additionally, they developed custom avatars to translate text messages back into sign language, adding a personal touch by animating the avatars to resemble the creators.
By combining real-time sign language recognition with avatar-based output, Eshara offers a comprehensive system that could revolutionise digital communication, making it more accessible for sign language users.
