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...

Descripción completa

Detalles Bibliográficos
Autores: Funosas, David, Sebastián-González, Esther, Morant, Jon, Marín Gómez, Oscar H., Mendoza, Irene, Mohedano Muñoz, Miguel A., Santamaría García, Eduardo, Bastianelli, Giulia, Márquez Rodríguez, Alba, Budka, Michał, Bota, Gerard, Alonso-Moya, Cristina D., Peña-Rubio, José M. de la, García de la Morena, Eladio L., Santa-Cruz, Manu, Nava, Pablo de la, Fernández-Tizón, Mario, Sánchez-Mateos, Hugo, Barrero, Adrián, Traba, Juan, Osiejuk, Tomasz S., Hart, Patrick J., Navine, Amanda K., Montoya Muñoz, Andrés F., Araújo, Carlos B. de, Rosa, Gabriel L. M., Torres, Ingrid M. D., Catalano, Ana L. C., Simões, Cassio Rachid, Llusia, Diego, Morales, Manuel B., Acebes, Pablo, Medina, Juan A., Brown, Nicholas, Astaras, Christos, Karmiris, Ilias, Navarrete, Elizabeth, Cauchoix, Maxime, Barbaro, Luc, Arend, Dominik, Müeller, Sandra, González-García, Fernando, González-Romero, Alberto, Mammides, Christos, Pontikis, Michaelangelo, Jacuzzi, Giordano, Olden, Julian D., Bombaci, Sara P., Marcacci, Gabriel, Jacot, Alain, Zurano, Juan P., Gangenova, Elena, Varela, Diego, Di Sallo, Facundo, Zurita, Gustavo A., Atemasov, Andrey, Tremblay, Junior A., Lamarre, Vincent, Hutschenreiter, Anja, Monroy-Ojeda, Alan, Díaz-Vallejo, Mauricio, Chaparro-Herrera, Sergio, Briers, Robert A., Sousa-Lima, Renata, Pinheiro, Thiago, Da Silva, Wigna C., Calvente, Alice, Paz, Raiane V., Salustio-Gomes, Carlos, Oliveira-Júnior, Dorgival D., Lima-Santos, Cicero S., Pichorim, Mauro, Molin, Anamaria Dal, Antonelli, Alexandre, Gogoleva, Svetlana, Palko, Igor, Trong, Hiếu V., Duarte, Marina H. L., Dos Santos Saturnino, Natalia, Silva, Samuel R., Rainho, Ana, Lopes, Paula, Schuchmann, Karl L., Marques, Marinêz I., De Oliverira Tissiani, Ana S., Littlewood, Nick A., Tuanmu, Mao Ning, Kepfer-Rojas, Sebastian, Aguilera, Andrea L., Brotons, Lluís, Feldman, Mariano J., Imbeau, Louis, Panwar, Pooja, Weed, Aaron S., Dehwal, Anant, Attisano, Alfredo, Theuerkauf, Jörn, Goodale, Eben, Darras, Kevin F. A., Pérez-Granados, Cristian
Tipo de recurso: artículo
Estado:Versión publicada
Fecha de publicación:2026
País:España
Institución:Consejo Superior de Investigaciones Científicas (CSIC)
Repositorio:DIGITAL.CSIC. Repositorio Institucional del CSIC
OAI Identifier:oai:digital.csic.es:10261/414382
Acceso en línea:http://hdl.handle.net/10261/414382
https://api.elsevier.com/content/abstract/scopus_id/105025359154
Access Level:acceso abierto
Palabra clave:Passive acoustic monitoring
Automated detection
Bird communities
BirdNET
Confidence threshold
Deep learning
Descripción
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.

Similares en LA Referencia