Camera-Based Vehicle Counting & Traffic Analytics System: Veyro

A virtual line is drawn on the camera image; vehicles crossing it are counted by direction (A→B and B→A). You can rename the directions to suit your road. Because every vehicle is tracked, it is counted once per line in the same direction.

Besides lines, you can define an occupancy zone: how many vehicles are in that zone at the same time is recorded every minute. Areas you do not want counted can be masked.

Veyro counts nine classes separately: cars, trucks, buses, motorcycles, bicycles, trains/trams, boats, aircraft and pedestrians. You decide which classes are counted.

You can measure counting accuracy yourself: in accuracy sessions you compare your manual count with the system's count. Because camera angle, light and image quality affect counting, we plan the camera placement together during setup.

The live view and wall display show vehicles with box and line overlays. The dashboard includes summaries, time series, breakdowns, a heat map and period comparisons; reports export as CSV, Excel and PDF and can be scheduled by e-mail daily, weekly or monthly.

Alerts are raised when a camera goes offline, a count threshold is exceeded, there is no traffic at all, or something anomalous happens; they can be delivered in the app, by e-mail and through signed webhooks. A read-only external API lets you pull counts into your own systems.

Veyro does not record video; frames are processed in memory only. It does not perform face or license plate recognition, and it counts pedestrians without identifying them. Optional snapshots are off by default. Retention periods for events, summaries and audit records are configurable.

Installation is on the customer's own server with Docker. RTSP/IP cameras, HTTP streams and video files can be used as sources. The panel supports Turkish and English with light and dark themes, role-based access, two-factor authentication (TOTP) and an audit log.