Accessible Scientific Graphs: Generating Alt Text for Scientific Graphs
This project addresses the long-standing accessibility gap between visual data representations — such as charts, graphs, and diagrams — and persons with visual impairements, who often cannot access this information through screen readers alone. In this project, we develop and evaluate methods for automatically generating meaningful, accurate descriptions of scientific and statistical graphics using AI, with the aim of making visual data genuinely accessible rather than requiring readers to rely on incomplete or inconsistent manual alt text. The project combines advances in multimodal AI with accessibility research, working toward practical tools and guidelines that can be adopted across scientific publishing, education, and the broader research community.
Supervisor
PhD Student
Project Timeline
September 2025 to December 2029
Funding
This project is funded by UZH DSI