Automatic License Plate Recognition based on vertical edge analysis

Our Take

Credit to Tomasz Myśliwiec for a clean, no-nonsense demo: real-time license plate recognition built with basic vertical edge detection in OpenCV. No cloud AI, no proprietary black box — just classic computer vision techniques that have been public knowledge for decades. That's exactly why this video is worth watching if you want to understand what Flock Safety and its competitors are actually selling.

Companies like Flock like to frame their cameras as some kind of advanced AI breakthrough, but plate-reading itself has never been the hard part. Edge detection, contour analysis, character segmentation — this is undergrad computer vision coursework. What Flock actually sells isn't the algorithm, it's the network: thousands of always-on cameras, a shared national database, and law enforcement access agreements that turn every driveway and intersection into a checkpoint. The tech in this video could run on a Raspberry Pi. The surveillance dragnet is a business and policy choice, not a technical inevitability.

That distinction matters when you're arguing with a city council or a neighborhood HOA about whether to install these cameras. The question was never 'can we read plates' — we've been able to do that cheaply for years. The question is who gets to store that data, for how long, and who they hand it to. Check our map to see what's already watching your streets, and hit up take action if you want to push back.