What's this about?
Jason Tucker repurposed his existing security cameras into an automatic bird identification system using BirdNET-Go.
What is BirdNET?
A Cornell Lab of Ornithology project that identifies bird species from audio — a neural network trained on thousands of species' calls, small enough to run on a Raspberry Pi. BirdNET-Go is a Go implementation for continuous local operation. The insight in this writeup is that security cameras already have microphones running continuously outdoors, which is exactly the sensor deployment you would otherwise have to build.
Why is it worth noticing?
HN 368 points, 95 comments
- It is the good version of the surveillance-hardware story that ran all week — the same cameras, pointed at something nobody objects to.
- Fully local, which is the property that makes always-on outdoor microphones acceptable at all.
- A clean example of ML that is small, specific, verifiable and useful, in a week dominated by the opposite.