LPR multicam working (license plate recognition)

Our Take

Credit where it's due: Rodrigo Dias built a working multi-camera license plate recognition pipeline — C++/OpenCV for detection, a custom Django API on the backend — as what looks like a personal or client project for something called "Stormsecurity." It's a solid technical demo. It's also a quiet warning.

We spend a lot of time here talking about Flock Safety specifically because they're the dominant player with the biggest camera network and the deepest law enforcement ties. But videos like this are a reminder that ALPR isn't some proprietary black box only a billion-dollar company can build. The core computer vision problem — spot a plate, read the characters, log it with a timestamp and location — is something one developer can stand up with open-source tools and a weekend of motivation. That means the fight against mass plate surveillance can't just be about regulating one vendor. It has to be about limiting what any camera, anywhere, is allowed to collect and retain, and who gets to see it.

If a hobbyist-grade system can track vehicle movements across multiple cameras, imagine what's already bolted to poles in your town. Check our map to see what's been documented near you, and if you want to push back on retention policies and data-sharing agreements before they become permanent, our take-action page has next steps.