Embedded Files
Building Generative Predictive Transformers for Sign Language
An EPSRC Programme Grant
1/5/2025-30/4/2030
Project Overview
Project Overview
SignGPT is a UKRI EPSRC Programme Grant focused on the development of an AI-powered translation system capable of unconstrained, bidirectional translation between British Sign Language (BSL) and English. The project will build the first generative predictive transformer for sign language, combining computer vision, sign linguistics, and machine learning. It is led by an interdisciplinary team from the University of Surrey, University of Oxford, and University College London, with direct involvement from Deaf organisations and community partners.
Meet the Team
Meet the Team
The SignGPT project includes:
University of Surrey (Lead): Prof. Richard Bowden – Computer Vision, AI
University of Oxford: Prof. Andrew Zisserman – Machine Learning, Large Models
UCL DCAL: Prof. Kearsy Cormier, Prof. Bencie Woll – BSL Linguistics
In addition, the project has an Advisory Board (AB) of academics, experts, and Deaf community leaders. The advisory board will provide external guidance and ethical oversight, identifying new academic and industry collaborative opportunities for data gathering, support connections with policy, education, and stakeholders and providing advice on research progress, dissemination and outreach. The AB will also give feedback on initial design at the start and on software tools before general release.
Confirmed organisations represented on the advisory board consist of BSLBT/LumoTV, Royal Association for Deaf People (RAD), British Deaf Association (BDA), Uni of Hamburg, Uni of Zurich, Gallaudet Uni, Microsoft Munich and BBC R & D.
Current Staff:
Current Staff:
Dr Steven Emery (SignGPT Project Manager) - s.emery@surrey.ac.uk
All partners are currently recruiting. Please reach out to us if you are interested in joining the team.
Questions?
Questions?
Contact r.bowden@surrey.ac.uk or s.emery@surrey.ac.uk to get more information on the project
Page updated
Google Sites
Report abuse