We build computer vision that helps blind and low-vision people, and anyone walking with divided attention, get around safely.
What we do
We focus on two groups: blind and low-vision (BLV) people, and pedestrians walking with divided attention, such as on their phones.
Eye-level images of real streets, labelled from a pedestrian’s point of view.
Compact models that spot obstacles fast enough to run on phones and wearables.
Turning detections into guidance that blind and low-vision people can rely on.
Research
We systematically review 108 peer-reviewed studies, published between 2008 and 2026, on assistive navigation technologies for blind and low-vision (BLV) individuals. We organize…
ROD-Dataset is a pedestrian-perspective benchmark for real-time obstacle detection. It contains more than 24,000 eye-level images and over 40,000 human-verified obstacle…
By the numbers
Our app and demos show how our detectors help people get around safely.
Paper announcements, datasets and code releases.