Transformer Models improve the acoustic recognition of buzz-pollinating bee species

Ecological Informatics, Journal Year: 2025, Volume and Issue: unknown, P. 103010 - 103010

Published: Jan. 1, 2025

Language: Английский

---

LRM-MVSR: A lightweight birdsong recognition model based on multi-view feature extraction enhancement and spatial relationship capture

Expert Systems with Applications, Journal Year: 2025, Volume and Issue: unknown, P. 126735 - 126735

Published: Feb. 1, 2025

Language: Английский

---

LEAVES: An open-source web-based tool for the scalable annotation and visualisation of large-scale ecoacoustic datasets using cluster analysis

Ecological Informatics, Journal Year: 2025, Volume and Issue: unknown, P. 103026 - 103026

Published: Feb. 1, 2025

Language: Английский

---

CNN-Based Optimization for Fish Species Classification: Tackling Environmental Variability, Class Imbalance, and Real-Time Constraints

Information, Journal Year: 2025, Volume and Issue: 16(2), P. 154 - 154

Published: Feb. 19, 2025

Automated fish species classification is essential for marine biodiversity monitoring, fisheries management, and ecological research. However, challenges such as environmental variability, class imbalance, computational demands hinder the development of robust models. This study investigates effectiveness convolutional neural network (CNN)-based models hybrid approaches to address these challenges. Eight CNN architectures, including DenseNet121, MobileNetV2, Xception, were compared alongside traditional classifiers like support vector machines (SVMs) random forest. DenseNet121 achieved highest accuracy (90.2%), leveraging its superior feature extraction generalization capabilities, while MobileNetV2 balanced (83.57%) with efficiency, processing images in 0.07 s, making it ideal real-time deployment. Advanced preprocessing techniques, data augmentation, turbidity simulation, transfer learning, employed enhance dataset robustness imbalance. Hybrid combining CNNs intermediate improved interpretability. Optimization pruning quantization, reduced model size by 73.7%, enabling deployment on resource-constrained devices. Grad-CAM visualizations further enhanced interpretability identifying key image regions influencing predictions. highlights potential CNN-based scalable, interpretable classification, offering actionable insights sustainable management conservation.

Language: Английский

---

Continental-scale behavioral response of birds to a total solar eclipse

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: April 8, 2025

Language: Английский

---

A scalable transfer learning workflow for extracting biological and behavioural insights from forest elephant vocalizations

Remote Sensing in Ecology and Conservation, Journal Year: 2025, Volume and Issue: unknown

Published: April 25, 2025

Abstract Animal vocalizations encode rich biological information—such as age, sex, behavioural context and emotional state—making bioacoustic analysis a promising non‐invasive method for assessing welfare population demography. However, traditional approaches, which rely on manually defined acoustic features, are time‐consuming, require specialized expertise may introduce subjective bias. These constraints reduce the feasibility of analysing increasingly large datasets generated by passive monitoring (PAM). Transfer learning with Convolutional Neural Networks (CNNs) offers scalable alternative enabling automatic feature extraction without predefined criteria. Here, we applied four pre‐trained CNNs—two general purpose models (VGGish YAMNet) two avian (Perch BirdNET)—to African forest elephant ( Loxodonta cyclotis ) recordings. We used dimensionality reduction algorithm (UMAP) to represent extracted features in dimensions evaluated these representations across three key tasks: (1) call‐type classification (rumble, roar trumpet), (2) rumble sub‐type identification (3) demographic analysis. A Random Forest classifier trained achieved near‐perfect accuracy rumbles, Perch attaining highest average (0.85) all call types. Clustering reduced identified biologically meaningful sub‐types—such adult female calls linked logistics—and provided clearer groupings than manual classification. Statistical analyses showed that factors including age significantly influenced variation P < 0.001), additional comparisons revealing clear differences among contexts (e.g. nursing, competition, separation), sexes multiple classes. BirdNET consistently outperformed when dealing complex or ambiguous calls. findings demonstrate transfer enables scalable, reproducible workflows capable detecting variation. Integrating this approach into PAM pipelines can enhance assessment dynamics, behaviour acoustically active species.

Language: Английский

---

Exploring the relationship between the soundscape and the environment: A systematic review

Ecological Indicators, Journal Year: 2024, Volume and Issue: 166, P. 112388 - 112388

Published: July 26, 2024

Language: Английский

---

Acoustic monitoring for tropical insect conservation

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: July 5, 2024

Abstract Monitoring the species-specific sounds produced by insects could provide us with a rapid, reliable, non-invasive measure of tropical ecosystem health and biodiversity. Although acoustic biodiversity monitoring has made rapid progress over past decade, focus been mostly on vertebrates, even though far outnumber them, soundscapes are dominated insect sounds. Here we an overview song features for major sound-producing groups, identify technological milestones describe impediments analyzing communities. We review some promising best-practices using singing profiling tracking diversity in rainforest ecosystems under threat. suggest roadmap joint research efforts to accelerate assessments based re-using wealth existing data from Passive Acoustic (PAM) combination curated multimedia repositories citizen science.

Language: Английский

---

The Australian fish chorus catalogue (2005–2023)

Frontiers in Remote Sensing, Journal Year: 2024, Volume and Issue: 5

Published: Dec. 12, 2024

Biological sources are significant contributors to aquatic soundscapes. Soniferous fish can dominate the soundscape in certain locations, at specific times and frequencies, particularly during production of choruses. Passive acoustic monitoring choruses provide important ecological information about soniferous populations. This study presents Australian Fish Chorus Catalogue, an inventory detected from 83 locations estuarine marine waters. The Catalogue contains data on chorus occurrence spectral temporal measurements, spectrographic images, audio examples 301 catalogue has been developed establish foundations ongoing effort document, quantify, compare, track We hope this open-access depository will be used as a reference for future research facilitate increase understanding choruses, which then applied management populations their respective ecosystems.

Language: Английский

---

Monitoring Postfire Biodiversity Dynamics in Mediterranean Pine Forests Using Acoustic Indices

Environments, Journal Year: 2024, Volume and Issue: 11(12), P. 277 - 277

Published: Dec. 4, 2024

In recent decades, climate change has significantly influenced the frequency and intensity of wildfires across Mediterranean pine forests. The loss forest cover can bring long-term ecological changes that impact overall biodiversity alter species composition. Understanding requires effective cost-efficient methods for monitoring postfire ecosystem dynamics. Passive acoustic (PAM) been increasingly used to monitor vocal at large spatial temporal scales. Using indices, where an area is inferred from structure soundscape, rather than more labor-intensive identification individual species, yielded mixed results, emphasizing importance testing their efficacy regional level. this study, we examined whether widely indicators were capturing in avifauna diversity Pinus halepensis stands with different fire burning histories (burnt 2001, 2009, 2018 unburnt >20 years) on Sithonia Peninsula, Greece. We recorded soundscape each stand using two–three sensors 11 days season March 2022 January 2023. calculated site following five indices: Acoustic Complexity Index (ACI), Diversity (ADI), Evenness (AEI), Normalized Difference Soundscape (NDSI), Bioacoustic (BI). Each index was then assessed terms its predicting local diversity, as estimated via two proxies—the richness (SR) Shannon (SDI) bird calls. Both SR SDI by having expert review calls detected within same dataset BirdNET convolutional neural network algorithm. A total 53 identified. Our analysis shows BI NDSI have highest potential dynamics propose development regional-scale observatories other fire-prone habitats, which will further improve our understanding how make best use indices a tool rapid assessments.

Language: Английский

---