According to Be My Eyes, a tech company that helps people with visual impairments, there are 340 million people around the world who are blind or have low vision. Today, that company announced a new collaboration with Microsoft that aims to help improve Microsoft's AI models to better serve those people.
In a new press release from Be My Eyes, the company stated it will be sending video data to Microsoft to help train its AI models. It added:
The video datasets represent the lived experience of the blind and low vision community and will be used to improve the accuracy and precision of scene understanding and descriptions, with the goal of increasing the utility of AI for the blind and low-vision community.
The company adds that only video data will be shared with Microsoft with this new agreement. Be My Eyes will also remove metadata from those videos that show user names, account names, and other personal information. Finally, the video data will not be used for any marketing or other services other than to train Microsoft's AI models.
In its own blog post about the new agreement, Microsoft offered up some more details on how it wants to use generative AI to help people with disabilities in general. Along with other companies like Apple, Google, Meta, and Amazon, Microsoft has been working with the University of Illinois Urbana-Champaign on its Speech Accessibility Project.
This project collects data to help AI models with speech recognition for people with disabilities, such as those who have ALS. It added:
This data helps AI models learn a broader understanding of the diversity of speech and expands scenarios where this technology can bring benefits – including accelerating non-standard speech data accuracy in mainstream speech platforms, like Azure Speech. The project is enhancing AI systems for people with disabilities by integrating diverse data and perspectives into the development process.
The press releases did not offer any financial details with this new partnership between Microsoft and Be My Eyes.
0 Comments - Add comment