Artificial intelligence — machine learning in particular — has advanced by leaps and bounds in recent years. A decade or two ago, machine learning was the domain of researchers and large corporations. Today, anyone with $50 in their bank account can tinker with powerful machine learning models. Object and face recognition machine learning models are some of the most common, because they're useful for a wide range of situations. To demonstrate Harlab's CM4Ext Nano board, Jeff Geerling built a leaf blower that starts up as soon as it sees his face.
Harlab's CM4Ext Nano is a carrier board for the Raspberry Pi Compute Module 4. The CM4 is very similar to a Raspberry Pi 4 Model B, but with all the ports and extraneous hardware omitted. It gives you the power and software compatibility of a normal Raspberry Pi board in a small, thin package that designers can slot into custom PCBs. The CM4Ext Nano reintroduces most of the ports that are missing from the CM4, but keeps everything sleek and compact. In this case, a Raspberry Pi 4 Model B would have worked just as well. But it is still nice to see the CM4 in use.
In addition to the CM4Ext Nano, this project called for a Raspberry Pi HQ camera, a lens, a tripod, a USB power bank, and an electric leaf blower. The Raspberry Pi can turn on the leaf blower by pushing its trigger using a hobby servo motor controlled through the GPIO pins. It does that when it detects a specific person — Geerling himself. OpenCV performs that detection using Google TensorFlow, which is a machine learning model. Geerling trained that model using many pictures of his own face. It only runs at 2-3 FPS (frames per second), but that is fast enough for it to recognize Geerling within a second or two and then activate the leaf blower.
It is unlikely that you will want to recreate this project exactly as it is, but it is a fantastic demonstration of how easy it is to get started with machine learning on a budget.