Open Collections

Zhao, Beiliang

University of British Columbia

Monitoring marine mammals through vocalizations provides scientists with essential insights into their abundance and distribution. Systems that classify these acoustic sensing from less than 10 to over 100000 hertz (Hz) enable precise surveillance of specific species, thereby playing a crucial role in the protection efforts for endangered species. However, current solutions based on traditional machine learning technologies face challenges in classifying a diverse range of marine mammal sounds with high accuracy. This study aims to improve classification accuracy for marine mammal vocalizations by refining the feature vectors from sounds, deploying advanced deep-learning models, and comparing the performance of various deep-learning models. Furthermore, this research utilizes the generalization capability of deep-learning models to process fresh acoustic data from different sources or additional species with high accuracy. The advancements proposed in this study offer a marked improvement in marine mammal surveillance, allowing for more precise species identification based on their vocal patterns.

Text

2024-01-08

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/87147

Electrical Engineering

University of British Columbia

UBCO

http://creativecommons.org/licenses/by-nc-nd/4.0/