YOLO KZ
Edge vision telemetry system converting roadside camera video into directional classification counts for vehicles, pedestrians, bicycles and scooters. Powered by an ASP.NET Core backend with PostgreSQL time-series storage and resilient idempotent batch ingestion.
Signal path
The working path from input to physical or measurable outcome.
Car · person · bicycle · scooter
Challenge & Requirements
Engineering constraints and goals.
Street activity had to be measured from a portable camera setup and converted into reliable, class-specific directional statistics.
Implementation
Architecture and engineering approach.
- 01
Used a phone on a tripod as a portable field camera for repeatable observation points.
- 02
Converted tracked line crossings into typed events for cars, people, bicycles and scooters.
- 03
Designed secure, idempotent batch ingestion so devices can reconnect without double-counting offline events.
- 04
Built the ASP.NET Core backend around projects, devices, rotating sessions, analytics and live counters.
Outcome & Deployment