what is lightbus fr?

LightBus is a student project, built by a five person team and awarded first place in a Microsoft AI academy competition, that helps visually impaired and cognitively impaired people confirm they are boarding the correct bus. It solves a specific everyday problem: even when a bus has arrived at a stop, a blind rider often cannot tell if it is the exact bus number they need. LightBus solves this using only a smartphone, without any beacons or extra hardware installed at bus stops. A user speaks the bus number or destination they want, and the app finds their location using GPS, looks up nearby bus stops and real time arrival information through South Korean public transit data APIs, then uses the phone's camera together with an object detection model to spot an approaching bus. Once a bus is detected, Azure's computer vision service reads the number plate through optical character recognition, and the app compares that number to what the user asked for before speaking the result back to them out loud. The system combines a React frontend built with the Web Speech and Geolocation browser APIs, a Node.js and Express backend, and a separate Python FastAPI service that runs the bus detection and plate recognition. The detection model is a YOLO based object detector retrained specifically to recognize buses, reaching roughly 90 percent precision and 80 percent recall in the team's testing, paired with OpenCV for image handling. Azure Machine Learning, Azure Computer Vision OCR, and Azure Speech Service for text to speech and speech to text round out the cloud side of the stack. The README describes future plans to expand coverage to bus routes nationwide rather than just Seoul, recommend the best stop to wait at based on a rider's destination, connect with smart glasses, and detect obstacles or hazards in real time. The project was built as part of a Microsoft AI School cohort and does not state an open source license.