Acoustic remote sensing with deep learning enables non-invasive estimation of seabird nest density

This repository contains the datasets, trained models, and code used in the study:

Terranova et al. (2026)

Acoustic remote sensing with deep learning enables non-invasive estimation of seabird nest density

This repository contains all scripts, trained models, and data products used for the automated detection of Ecstatic Display Songs (EDS) in African penguins, as well as their application to long-duration acoustic recordings and subsequent nest density analyses.

├── data/
│   ├── cnn_vs_manual_annotation.tsv
│   ├── eds_cnn_detections_stp2024_filtered.tsv
│   ├── eds_cnn_detections_stp2025_filtered.tsv
│   ├── eds_peak_and_nest_counts_by_point_2024_2025.csv
│   ├── merged_eds_weather_2024.csv
│   ├── merged_eds_weather_2025.csv
│   ├── nest_count_2024.csv
│   ├── nest_count_2025.csv
│
│
│── model/
│   ├── best_model.h5
│   ├── config.json
│   ├── grid_search_results.csv
│   ├── history.pkl
│   ├── summary.txt
│   
│
├── code/
│   ├── python/
│   │   ├── cnn_vs_manual_annotation.ipynb
│   │   ├── eds_cnn_dataset_and_training.py
│   │   ├── eds_cnn_inference_long_recordings_stp2024.py
│   │   ├── eds_cnn_inference_long_recordings_stp2025.py
│   │   └── requirements.txt
│   └── R/
│       ├── eds_nest_density_gam_analysis.R
│       
└── metadata_readme.txt

Due to file size constraints, the raw acoustic recordings used in this study are not included in this repository but are available from the authors upon request.