North Atlantic right whale

right whale
A spectrogram based model for detecting North Atlantic right whale up-sweeps

Introduction

This is a spectrogram based model for detecting right whale up-sweeps developed by Shiu and Palmer et al. (2020). The model was tested on a standardised dataset from the DCLDE conference in 2013 (Detection, Classification, Localisation, and Density Estimation)to quantify the performance compared to other leading automated analysis approaches. Shiu/Palmer et al.’s 2020 deep learning approach outperformed all those efforts by a very wide margin (see Figure 1 in their paper) and was an early demonstration of the potential of deep learning approached in marine bioacoustics.

Note that this model is also available as an example model within PAMGuard.

The Shiu and Palmer right whale classifier running in PAMGuard

Species

  1. North Atlantic right whale (Eubalaena glacialis)

Geographic location of training data

Cape Cod Bay

Reference

Shiu, Y., Palmer, K.J., Roch, M.A., Fleishman, E., Liu, X., Nosal, E.-M., Helble, T., Cholewiak, D., Gillespie, D., Klinck, H., 2020. Deep neural networks for automated detection of marine mammal species. Scientific Reports 10, 607. https://doi.org/10.1038/s41598-020-57549-y

Model

The model can be found as part of the tutorial data set along with example wav files DOI

An example settings file can be found here.