Exciting New Publication by Dr. Gui Frainer on detecting and identifying endangered Indian Ocean Humpback Dolphin vocalizations.

We're thrilled to share the latest research paper led by Dr. Gui Frainer that could be a game-changer for the conservation of the endangered Indian Ocean humpback dolphin (Sousa plumbea) in South African waters!
Our research group developed an innovative framework using Convolutional Neural Networks (CNNs) to detect and identify dolphin vocalizations, paving the way for more effective passive acoustic monitoring.
Title: Automatic detection and taxonomic identification of dolphin vocalisations using convolutional neural networks for passive acoustic monitoring

https://www.sciencedirect.com/.../pii/S1574954123003205...

Here are some key findings:

Detection Model: The best model achieved an impressive accuracy of 84.4% for all dolphin vocalizations and 89.5% for high signal-to-noise ratio vocalizations.

Identification Model: This model correctly identified Indian Ocean humpback dolphins (96.9%), Indo-Pacific bottlenose dolphins (100%), and common dolphins (78%) encounters in the testing dataset.
This breakthrough research not only helps in the conservation of the endangered Indian Ocean humpback dolphins but also offers valuable insights into the development of open-source tools for studying other species and populations.

You can find out more about the new tool here (https://github.com/Gui-Frainer/CetusID)
This research was supported by National Research Foundation and Microsoft Azure through the 'AI for Earth' program.

Julkaistu lokakuu 16, 2023 03:31 IP. käyttäjältä seasearch