Tegwyn Twmffat Turns to Deep Learning for a Smart Species-Identifying Bat Detector Build

Using machine learning to recognize each species, the detector can note bat types and upload data to the cloud over LoRa.

Tegwyn Twmffat has published a build for an edge-AI bat detector built around an NVIDIA Jetson Nano module or Raspberry Pi 4, taking advantage of its computing power to run a machine learning system to identify bat species' via their unique ultrasonic chirps.

"Initially I started off using a package designed for music classification called 'PyAudioAnalysis' which gave options for both Random Forest and then human voice recognition Deep Learning using TensorFlow," Twmffat explains. "Both systems worked OK, but the results were very poor. After some time chatting on this very friendly Facebook group — Bat Call Sound Analysis Workshop — I found a software package written in the R language with a decent tutorial that worked well within a few hours of tweaking."

The software runs on an NVIDIA Jetson Nano Development Kit or a Raspberry Pi 4 Model B 4GB, and includes a custom user interface built on GTK 3 running on an 800x480 touch-screen monitor connected over HDMI. A 12V battery pack, monitored through an analogue-to-digital converter (ADC) add-on, keeps everything running, while an UltraMic 384 captures the audio - and the whole lot is housed in a waterproof case for field use. There's also the option of transmitting data to a remote server over LoRa, though Twmffat notes this is currently only functional on the Raspberry Pi 4 version of the build.

The biggest trick of the project was in finding data to feed the recognition system. "Finding quality bat data for my country [was a challenge]," Twmffat notes. "In theory, there should be numerous databases of full spectrum audio recordings in the UK and France, but when actually trying to download audio files, most of them seem to have been closed down or limited to the more obscure 'social calls.'

"The only option was to make my own recordings which was actually great fun and I managed to find 6 species of bat in my back yard. This was enough to get going."

More details can be found on Twmfatt's project page. Those working on other NVIDIA Jetson AI-powered projects, meanwhile, have just a few days left to submit their entries into the Hackster.io AI at the Edge Challenge.

Gareth Halfacree
Freelance journalist, technical author, hacker, tinkerer, erstwhile sysadmin. For hire: freelance@halfacree.co.uk.
Latest articles
Sponsored articles
Related articles
Latest articles
Read more
Related articles