This summer I took a 6-week rapid prototyping class. We met once a week and played with sensors hooked up to an Arduino or a Raspberry Pi. For the final project we were asked to prototype a censor that could be used to understand something about city operations.
We made a rat detector that could be mounted somewhere on the subway stations to count rats. We weren't too concerned with operational details and costs. It would probably be prohibitively expensive to deploy enough of them to collect a good census of the rat population.
Juan Medina wrote image processing in OpenCV and Python. Emil Christensen wrote code to run the camera and the OpenCV pipeline in parallel. I went hunting for rats in New York subway with our makeshift rig and did not get arrested.
The detector is really just a moving object detector - it does not classify objects as rats.
- Raspberry Pi 2
- Raspberry Pi camera module (with removed infrared filter)
- Power bank
- Breadboard with a LED camera indicator
The code on GitHub - it's a mess.