A Cloud and Vision-Based Navigation System for Blind People
Software as a Service
NoCategory
AI & Machine LearningClients
Techstack
Purpose
A blind person moving through an unfamiliar space needs to know what is in the way before they reach it. A research team set out to build an aid that could recognize obstacles in real time and, just as importantly, keep its readings and status available in the cloud rather than trapped on the device.
Description
We built a camera-based prototype that runs live object detection to identify obstacles in the user's path, testing several vision models across design iterations to find the one that held up best. Every detection syncs to the cloud so the system's data and state can be watched from elsewhere, with a desktop interface for configuration and review. It is an assistive computer-vision problem with a real cloud backbone — the kind of build we take on and ship.
