There’s a stop sign outside [Devin Gaffney]’s house that, apparently, no one actually stops at. In order to avoid the traffic and delays on a major thoroughfare, cars take the road behind [Devin Gaffney]’s house, but he noticed a lot of cars didn’t bother to stop at the stop sign. He had a Raspberry Pi and a camera, so he set them up to detect the violating cars.
His setup is pretty standard – Raspberry Pi and camera pointed outside at the intersection. He’s running OpenCV and using machine learning to detect the cars and determine if they have run the stop sign or not. His website has some nice charts showing when the violations occurred by hour and by day of the week. Also on the site are links that you can use to help train the system in noticing cars, cars that run the stop sign, determining if there’s enough of the video to determine if there’s a violation, and whether or not there’s a car going the wrong way through the intersection.
This is an interesting use of the Pi and OpenCV; there’s no guarantee that this will help the people of [Devin Gaffney]’s neighborhood, but hopefully gives them some ammunition (assuming they want something done about the intersection.) It’s a cheap and easy setup and it’s nice to let the community have a hand in training the system. For more OpenCV, check out this article on taking the perfect jump shot or this one which tries to quantify cloudiness. Cool stuff.