Privacy statement

RoadSense operates TrafficAI, a project that analyzes publicly available camera streams. This page states what data the system handles and what it does not.

What the system observes

TrafficAI processes video from public traffic cameras operated by the Virginia Department of Transportation. It detects object classes and tracks: vehicles, pedestrians, and cyclists. It records counts, distances, speeds, timings, and derived risk scores.

What the system does not do

• It performs no facial recognition and holds no facial data.
• It reads no license plates and stores no plate text.
• It cannot follow a vehicle from one camera to another. A track identifier exists only within a single camera's field of view and is discarded when the vehicle leaves the frame.
• It holds no personal data, and no record it produces could identify an individual person or vehicle.

Scene descriptions elsewhere on this site are published exactly as the vision language model wrote them, and the model sometimes remarks on what is visible in a frame, including that a vehicle has a plate. Nothing it says is extracted, parsed or kept.

Video handling and retention

Frames are processed in memory and discarded. The system writes no video archive. Annotated snapshots are saved only for alert evidence and human review, and that setting is disabled by default. Where cloud storage is used, lifecycle rules delete raw event records after seven days and incident records after ninety.

This website

This site is a static set of files. It sets no cookies, runs no analytics, and collects nothing about visitors. Map tiles, when the browser can reach them, are requested from OpenStreetMap, whose own privacy terms apply to that request.

Contact

Questions about this statement can be sent to vihaansethi1@gmail.com.