Biology subjects

Cretois, B.

Publications and source records attributed to Cretois, B..

4 recordsLinked to original sources

bioRxiv2025

Data-driven Sampling Strategies for Fine-Tuning Bird Detection Models

Passive Acoustic Monitoring 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, PAM collects vast amounts of data, and annotating bird vocalizations requires specialized expetise. 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 passive acoustic monitoring.

bioinformatics↗

bioRxiv2024

National-scale acoustic monitoring of avian biodiversity and phenology

Billions of birds migrate annually, triggered by endogenous behaviors but also by ecoclimatic drivers which are shifting with climate change. These dynamics play out over huge spatiotemporal scales, making monitoring of phenology challenging with traditional biodiversity survey approaches. In this study, over a complete spring migration season (April through June), we collected 37,429 hours of audio from 28 networked sensors in forests across Norway using a nationwide passive acoustic monitoring (PAM) system. We applied an open-source detection algorithm to automatically classify bird vocalizations; through expert validation we found the algorithm classified 57 species (14 full migrants) with at least 80% precision. Using these automated detections, we developed regional arrival curves for three common migratory passerines: Willow Warbler, Common Chiffchaff, and Spotted Flycatcher. We then demonstrate that PAM detections can be used to train audio species distribution models that map how species vocalization probability changes across Norway during spring migration. Lastly, we discuss how PAM can complement existing manual surveys to support the design and implementation of effective policy and conservation measures.

ecology↗

bioRxiv2023

Snowmobile noise alters bird vocalization patterns during winter and pre-breeding season

Noise pollution poses a significant threat to ecosystems worldwide, disrupting animal communication and causing cascading effects on biodiversity. In this study, we focus on the impact of snowmobile noise on avian vocalizations during the non-breeding winter season, a less-studied area in soundscape ecology. We developed a pipeline relying on deep learning methods to detect snowmobile noise and applied it to a large acoustic monitoring dataset collected in Yellowstone National Park. Our results demonstrate the effectiveness of the snowmobile detection model in identifying snowmobile noise and reveal an association between snowmobile passage and changes in avian vocalization patterns. Snowmobile noise led to a decrease in the frequency of bird vocalizations during mornings and evenings, potentially affecting winter and pre-breeding behaviors such as foraging, predator avoidance and successfully finding a mate. However, we observed a recovery in avian vocalizations after detection of snowmobiles during mornings and afternoons, indicating some resilience to sporadic noise events. These findings emphasize the need to consider noise impacts in the non-breeding season and provide valuable insights for natural resource managers to minimize disturbance and protect critical avian habitats. The deep learning approach presented in this study offers an efficient and accurate means of analyzing large-scale acoustic monitoring data and contributes to a comprehensive understanding of the cumulative impacts of multiple stressors on avian communities. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=88 SRC="FIGDIR/small/548680v1_ufig1.gif" ALT="Figure 1"> View larger version (43K): org.highwire.dtl.DTLVardef@1deda15org.highwire.dtl.DTLVardef@1bfa2e4org.highwire.dtl.DTLVardef@5f2cc2org.highwire.dtl.DTLVardef@739ecd_HPS_FORMAT_FIGEXP M_FIG C_FIG

ecology↗

bioRxiv2022

Automated speech detection in eco-acoustic data enables privacy protection and human disturbance quantification

Eco-acoustic monitoring is increasingly being used to map biodiversity across huge scales, yet little thought is given to the privacy concerns and potential scientific value of inadvertently recorded human speech. Automated speech detection is possible using Voice Activity Detection (VAD) models, but existing approaches have been developed for indoor or urban use cases, rather than diverse natural soundscapes. In this study we used a data augmentation approach to create ecoVAD, a convolutional neural network designed for robust voice detection in eco-acoustic data. We performed playback experiments using speech samples from a woman, man, and child in two ecosystems in Borsa, Norway, and showed that ecoVAD was able to accurately detect voices at distances of up to 20m - at which point the speech was unintelligible. We compared ecoVAD with two existing VAD models and found that ecoVAD consistently outperformed the state-of-the-art (mean F1 scores: ecoVAD, 0.917; pyannote, 0.890; WebRTC VAD, 0.876). Using long-term passive recordings from a popular hiking location in Bymarka, Norway, we found that the frequency of speech detections was linked closely to peak traffic hours (using bus timings) demonstrating how VAD models can be used to quantify human activity with a fine temporal resolution. Anonymising audio data effectively using VAD models will allow eco-acoustic monitoring to continue to deliver invaluable ecological insight at scale, whilst minimising the risk of data misuse. Furthermore, using speech detections as a fine scale measure of human disturbance opens new possibilities for studying subtle human-wildlife interactions on the vast scales made possible by eco-acoustic monitoring technology. Author SummaryEco-acoustic monitoring (i.e., recording and analysing the sounds of an ecosystem) allows us to monitor biodiversity and ecological health on vast spatial and temporal scales. However, as networks of audio recorders are increasingly being deployed across the world by scientists and land managers, little thought has been given to the implications of inadvertently recorded human speech in these datasets. We developed a neural-network model to detect and remove speech from audio data, and showed it performed more accurately than existing state-of-the-art models when evaluated on eco-acoustic datasets. Furthermore, we showed frequency of voice detections could be used to monitor human pressure on an ecosystem. Our work paves the way for ethical and responsible processing of eco-acoustic data and provides a novel approach to studying human wildlife interactions on a large scale with high temporal resolution.

bioinformatics↗