Listening to the bush: How AI can help NZ rid its wildnerness of ‘hold-out’ pest possums
AI listening devices could help find New Zealand’s last remaining possums – but only if they don’t mistake bird calls for pests. Here’s how a new fix works.
New Zealand's wildlife conservation efforts are facing the challenge of removing the last few remaining possums from wilderness areas. Detecting these elusive animals is critical to the Predator Free 2050 programme, but traditional methods can be laborious and expensive. One promising solution involves using microphones that record nighttime sounds and artificial intelligence (AI) software to scan the thousands of hours of audio for possum calls.
Our recent research shows that this approach can be effective, but it requires careful training to avoid false alarms. When AI incorrectly identifies other animal sounds as possum calls, conservation workers can waste time traveling to remote locations where the animals aren't actually present. To address this issue, we developed a new training method called "cross-model confusion mapping."
Rather than directly asking the AI to identify possum sounds, we first used BirdNET, an AI system trained to recognize over 6,000 bird species. BirdNET can classify sounds as birds, which helps reveal which bird species are most likely to be mistaken for possums. By adding recordings of these birds to our training data as "hard negatives," we teach the AI to distinguish between possum calls and these other bird sounds.
We tested the retrained AI models on various recordings from New Zealand forests, including natural sounds, bird calls, insects, rustling vegetation, wind, rain, and possum vocalizations. The results were promising. Models trained using our approach produced far fewer false alarms while maintaining high detection accuracy. This was particularly important when the models were tested on completely new recordings they had never encountered before.
While the approach worked well for possums, it could potentially be adapted to detect other invasive pest species like stoats and rats. However, further field testing is needed before the system can be widely deployed. Nonetheless, as bioacoustic monitoring becomes more common, AI could become a valuable conservation tool in New Zealand's efforts to rid the country of pests by 2050.
Written by urgent.news from The Conversation AU's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.