A global assessment of BirdNET performance: Differences among continents, biomes, and species
Recent advances in machine learning have accelerated automated species detection across diverse ecological domains, enabling large-scale, non-invasive monitoring of biodiversity. In ornithological research, the combination of passive acoustic monitoring (PAM) and rapidly-developing novel identificat...
| Autores: | , , , , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión publicada |
| Fecha de publicación: | 2026 |
| País: | España |
| Institución: | Universidad de Sevilla (US) |
| Repositorio: | idUS. Depósito de Investigación de la Universidad de Sevilla |
| OAI Identifier: | oai:idus.us.es:11441/182509 |
| Acceso en línea: | https://hdl.handle.net/11441/182509 https://doi.org/10.1016/j.ecolind.2025.114550 |
| Access Level: | acceso abierto |
| Palabra clave: | Passive acoustic monitoring Bird communities BirdNET Deep learning Automated detection Confidence threshold |
| Sumario: | Recent advances in machine learning have accelerated automated species detection across diverse ecological domains, enabling large-scale, non-invasive monitoring of biodiversity. In ornithological research, the combination of passive acoustic monitoring (PAM) and rapidly-developing novel identification tools such as BirdNET—a deep learning–based sound recognition algorithm—offers new opportunities for surveying vocally active bird communities. Here, we present the first worldwide evaluation of BirdNET using 4224 one-minute recordings from 67 sites across all continents annotated by local experts. More specifically, we assessed the capacity of BirdNET to accurately identify individual vocalizations and characterize bird communities based on the automated analysis of passively collected soundscapes. We further analyzed how its performance varies across continents, biomes, species, and minimum confidence thresholds. The proportion of correct BirdNET predictions (precision) was generally high and consistent across continents (range: 0.57–0.71) and biomes (range: 0.55–0.76). In contrast, the proportion of vocalizations successfully detected (recall) was generally lower and more heterogeneous across continents (range: 0.24–0.52) and biomes (range: 0.34–0.72), reflecting differences in species coverage and local ecological context. BirdNET predictive power, as measured by the Precision-Recall Area Under the Curve (PR AUC; higher values indicating better performance), was highest in North America, Oceania, and Europe (range: 0.16–0.23), moderate in Central/South America (0.13), and lowest in Africa and Asia (range: 0.03–0.04). Species-specific analyses revealed substantial heterogeneity in detection accuracy, with optimal confidence thresholds varying widely by species and analytical goal. Our results establish a global reference point for BirdNET reliability and highlight where algorithmic refinement and expanded acoustic sampling are most needed. |
|---|