| Title: |
Automatic Detection and Taxonomic Identification of Dolphin Vocalisations using Convolutional Neural Networks for Passive Acoustic Monitoring |
| Authors: |
Guilherme Frainer; Emmanuel Dufourq; Jack Fearey; Sasha Dines; Rachel Probert; Simon Elwen; Tess Gridley |
| Publisher Information: |
Zenodo |
| Publication Year: |
2023 |
| Collection: |
Zenodo |
| Subject Terms: |
Convolutional neural networks; endangered species; Indian Ocean humpback dolphin; machine learning; passive acoustic monitoring; sound detection; species identification |
| Description: |
A novel framework for evaluating dolphin sound detection and species identification is proposed to aid passive acoustic monitoring studies on the endangered Indian Ocean humpback dolphin (Sousa plumbea) in South African waters. A multi-class classifier that encompasses all the dolphin species occurring in South African waters has yet to be developed. Convolutional Neural Networks (CNNs) were used for both detection and identification tasks, and their performance were evaluated using custom and pre-trained architectures (transfer learning). The developed framework assists in finding suitable CNN hyper-parameters for classification tasks on complex dolphin sounds and can be easily adapted for other species or populations. The application developed here may assist in future long-term studies of endangered species living in highly diverse habitats using passive acoustic monitoring. We provide a subset of the acoustic recordings used for the demonstration notebook (CetusID_Demo), including labels for the training data, as well as the testing dataset for both models. The detection model was evaluated by comparing ground-truth data with the model's performance on one entire day of recording (TestingData_Mooring1). Recordings from moored hydrophones (or drifting buoy, for Delphinus delphis) with known species identification were used to test the species identification model (TestingData_SpeciesName). |
| Document Type: |
article in journal/newspaper |
| Language: |
unknown |
| Relation: |
https://zenodo.org/records/8074949; oai:zenodo.org:8074949; https://doi.org/10.5281/zenodo.8074949 |
| DOI: |
10.5281/zenodo.8074949 |
| Availability: |
https://doi.org/10.5281/zenodo.8074949; https://zenodo.org/records/8074949 |
| Rights: |
Creative Commons Attribution 4.0 International ; cc-by-4.0 ; https://creativecommons.org/licenses/by/4.0/legalcode |
| Accession Number: |
edsbas.6C78431A |
| Database: |
BASE |