Data-driven sampling strategies for fine-tuning bird detection models
The Journal of the Acoustical Society of America · 2026
DOI: 10.1121/10.0043947 ↗Kurz gefasst
Fachartikel in der Zeitschrift The Journal of the Acoustical Society of America, 2026. Verfasst von Corentin Bernard und 5 weiteren Autor:innen. Beteiligte Einrichtung: Univ Toulon, Aix Marseille Univ, CNRS, LIS, Toulon, France. Hauptthemen laut MEDLINE: Vocalization, Animal, Acoustics, Birds, Signal Processing, Computer-Assisted. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 59 Literaturangaben. Gefördert von: OFB, ANR, Biodiversa+.
Abstract (Originalsprache)
Passive acoustic monitoring (PAM) has emerged as a promising tool for collecting ecological data, particularly in the context of bird population monitoring. Bird species can be automatically identified using pre-trained models, such as BirdNET. The performance of these models can be significantly improved through fine-tuning with annotated samples recorded in the specific acoustic conditions in which the microphones are deployed. However, passive acoustic monitoring (PAM) collects vast amounts of data, and annotating bird vocalizations requires specialized expertise. As a result, only a very small portion of the recordings can be effectively labeled. Selecting the most relevant samples to annotate in order to maximize performance in model fine-tuning remains a significant challenge. First, a regularization technique addresses the challenge of class imbalance during model fine-tuning. Next, a data-driven methodology is developed, introducing the influence score, which quantifies the impact of individual training samples on model performance to inform sampling strategies. A linear model is proposed to estimate the influence score for generalization to unseen data. Finally, several sampling strategies are compared, based on acoustic indices and predictions of the pre-trained model. Together, these contributions enable the identification of efficient annotation strategies to overcome the challenges of limited annotation resources in large-scale PAM.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- Univ Toulon, Aix Marseille Univ, CNRS, LIS, Toulon, France.
- Univ Toulon, CIAN, Toulon, France.
- Department of Intelligent Systems, Tilburg University, Tilburg, The Netherlands.
- Norwegian Institute for Nature Research, Trondheim, Norway.
- Naturalis Biodiversity Center, Leiden, The Netherlands.
- CNRS, ILLS, Montreal, Canada.
ORCID-Profile: Corentin Bernard, Ben McEwen, Benjamin Cretois, Hervé Glotin, Dan Stowell, Ricard Marxer
Förderung
- OFB
- ANR
- Biodiversa+
Zitieren
APA
Bernard, C., McEwen, B., Cretois, B., Glotin, H., Stowell, D., & Marxer, R. (2026). Data-driven sampling strategies for fine-tuning bird detection models. The Journal of the Acoustical Society of America. https://doi.org/10.1121/10.0043947
Vancouver
Bernard C, McEwen B, Cretois B, Glotin H, Stowell D, Marxer R. Data-driven sampling strategies for fine-tuning bird detection models. The Journal of the Acoustical Society of America. 2026. doi:10.1121/10.0043947
BibTeX
@article{Bernard2026,
author = {Bernard, Corentin and McEwen, Ben and Cretois, Benjamin and Glotin, Hervé and Stowell, Dan and Marxer, Ricard},
title = {Data-driven sampling strategies for fine-tuning bird detection models},
journal = {The Journal of the Acoustical Society of America},
year = {2026},
doi = {10.1121/10.0043947}
}
EUVIMED