An artificial intelligence recognition algorithm for Yangtze finless porpoise

Processing marine-mammal signals for species classification and monitoring of endangered marine mammals are problems that have recently attracted attention in the scientific literature. Currently, the detection of signals of interest is typically accomplished through a combination of visual inspection of spectrograms and listening to the data. This paper presented an automatic identification algorithm for the Yangtze finless porpoise based on Hilbert Huang Transform and BP artificial neural network. The algorithm includes three steps: signal preprocessing, feature extraction and signal identification. In feature extraction stage of the algorithm, the algorithm extracts a 11-Dimension signal feature vector based on Hilbert Huang transform, Shannon entropy and Fourier transform. In the identification stage, the BP artificial neural network is trained by using the feature vector as input. At last, some experimental acoustic data files of finless porpoise are used to test the validity of the automatic identification algorithm. The identification rate of the algorithm proposed in this paper reaches 90% with highest false positive rate (<;92 per hour) according to the human observation on the time-frequency spectrum. Because the Yangtze finless porpoise is one of the most critically endangered mammals in the world, so the presented method has great practical significance for protecting and monitoring the Yangtze finless porpoise in the wild.