News

  1. News
  2. 2020
  3. October
  4. ORBIT project working with blind and low vision people enters worldwide phase
News from City, University of London
AI4A photo1hero
Science & Technology Series: Announcements

ORBIT project working with blind and low vision people enters worldwide phase

The City-led project, comprising researchers from City, Microsoft Research and University of Oxford, funded by Microsoft AI for Accessibility, is ready to invite more participants, and will collect an even larger dataset from blind and low vision people from around the globe.
by John Stevenson (Senior Communications Officer)

The City-led ORBIT project, which has now entered its second phase, is working with blind and low vision people worldwide to build new Artificial Intelligence (AI) technology that helps to identify personal items.

566676Misplacing a bunch of keys can be annoying for the sighted, but it is even more complicated for the visually challenged. For a blind person it can frequently be a question of ‘how can you quickly find your keys and not mistake them for your housemate’s keys’?

Novel area

The ORBIT research project, comprising researchers from City, Microsoft Research and University of Oxford and funded by Microsoft AI for Accessibility, is working with blind and low vision users to record videos of objects that are important to them.

The collected data will enable the team to develop personal object recognisers, a novel area of AI and computer vision.

Current applications like Seeing AI or TapTapSee, for blind and low vision users, only identify common items, but not those which are unique to a user.

This research aims to fill this gap but AI only works if there is enough data.

As project lead, Dr Simone Stumpf, says:

Data is the fuel that AI runs on. Currently we don’t have any data that we can use from blind and low vision people to develop these personal object recognisers. Our project is a double-first. We are first to collect a large data set from people who are blind and low vision, and first to get them to record videos for this.

In a first phase of data collection, the team has already collected a dataset for nearly 400 objects in over 3000 videos from 48 users in the UK.

Dr Lida Theodorou, working as a researcher on this project, says:

“There were a few common objects including white canes, keys, wallets, earphones and backpacks but we were really surprised by how individual personal objects were. We even had a face mask and a wheelie bin! We expect that this range will increase even more, featuring some really unusual items, once we collect data from people all over the world.”

The team are now ready to invite more participants, collecting an even larger dataset from blind and low vision people from around the globe.

To contribute to this project, you will need to be over 18 years old, be blind or have low vision, and have an iPhone or iPad. To find out more about this project, including how to take part, please visit this weblink.

Please contact Dr Lida Theodorou at Lida.Theodorou.2@city.ac.uk with any questions relevant to the project.

Tags , , , , ,
Join the conversation #DrSimoneStumpf#ORBIT#DrLidaTheodorou#DepartmentofComputerScience#HCID#MicrosoftAIforAccessibility
Share this article