Make: Jumpstarting Raspberry Pi Vision: Machine Learning and Facial Recognition on a Single-Board Computer

If a camera takes a picture and no one notices, did it really happen? In this book, we’re going to show you how to make a working Raspberry Pi–based camera system so that you can capture time-lapse images and view via WiFi, trigger the camera if motion is detected, and even carry out basic facial recognition as an introduction to machine learning methods. Take a Raspberry Pi and add a camera module, and you have a programmable camera. Add some software, and you can start to do interesting surveillance and automatic object recognition work with it. Activate the Pi as a WiFi node and you can do all these wonderful things from a distance. A good surveillance system does more than take pictures. It should also turn those pictures into actionable information that increases your knowledge. That’s now easily done in software, and we’re going to show you how. This book pulls together a set of little tricks—setting up Pi cameras, making a Pi broadcast as a WiFi device, adding time lapse and motion detection and face recognition, and sticking a battery pack on it so it can function anywhere—to create portable spy cameras. We’ve used these rigs in everything from “Find the Pi” party contests, to implementing privacy-respecting security in our lab, to showing off modern tech like facial recognition. Other uses could include monitoring deer and wildlife, checking your house mailbox for mail arrival, and capturing time-lapse sequences of natural events or traffic patterns.