Detection of North Atlantic Right Whales with a Hybrid System of CNN and Dictionary Learning

In this paper, a hybrid approach of using Convolutional Neural Network (CNN) and dictionary learning is proposed for the detection of North Atlantic Right Whale (NARW) up-calls. The CNN was applied to extract features from NARW up-calls after these sound signals are converted into spectrograms. The features were then used to train a dictionary with a dictionary learning algorithm. The hybrid system of CNN and dictionary learning was compared with other methods using empirical data. It is shown in the paper that the proposed hybrid system produces superior results for detecting NARW up-calls.