This project leverages a BirdNET-Pi station I set up late last year. This digital counter pulls the latest daily count and species heard by the microphone in my backyard.
The most common species to visit and chirp is the House Finch. This bird blows any other species out of the water noise-wise.
I made a write-up on the project and how I programmed everything (and if you want to see a video of it).
That is so cool. How well does it classify sounds as actual birds? If someone tries to imitate one would it count?
It works extremely well! You can’t fool it unless you play a bird sound from your phone. You can set a defined confidence threshold to your liking. I usually leave mine at 50%. If you’ve ever used Cornell’s Merlin app (like Shazam for birds) it uses the same model and works just as well.
Wow a useful neural network on a PI?! That’s awesome!
Yep! It’s great! It is a super fun project for the tech-savvy tinkerer. If you are a more casual bird lover, I would recommend something like a Haikubox. It uses the same BirdNET neural network but is plug-and-play. I did a writeup about it on my blog.
thats so cool!
