Automated Bioacoustic Monitoring for South African Bird Species on Unlabeled Data
This work addresses the challenge of adapting passive acoustic monitoring to new bird species and habitats for conservation, though it is incremental as it builds on existing SED methods.
The researchers tackled the problem of time-consuming and noisy biodiversity monitoring by developing a framework that automatically extracts labeled data for bird species from platforms, trains CRNN models, and achieves an F1 score of 0.73 on unprocessed real-world recordings in urban habitats.
Analyses for biodiversity monitoring based on passive acoustic monitoring (PAM) recordings is time-consuming and challenged by the presence of background noise in recordings. Existing models for sound event detection (SED) worked only on certain avian species and the development of further models required labeled data. The developed framework automatically extracted labeled data from available platforms for selected avian species. The labeled data were embedded into recordings, including environmental sounds and noise, and were used to train convolutional recurrent neural network (CRNN) models. The models were evaluated on unprocessed real world data recorded in urban KwaZulu-Natal habitats. The Adapted SED-CRNN model reached a F1 score of 0.73, demonstrating its efficiency under noisy, real-world conditions. The proposed approach to automatically extract labeled data for chosen avian species enables an easy adaption of PAM to other species and habitats for future conservation projects.