Age-groups classification of Irrawaddy dolphins based on dorsal fin geometric morphological features

Abstract

The Irrawaddy dolphin is a species at the top of the food chain, but its population is currently endangered. Hence, the critical need to take effective conservation measures for Irrawaddy dolphins, particularly in population reproduction. The age structure of Irrawaddy dolphins can be studied to understand the population reproduction status and provide reference for formulating scientific conservation measures. Unlike age estimation methods based on invasive biological tissue detection and individual biological target detection methods based on audio features or image pixel features, this paper proposes the method to extract the Irrawaddy dolphin's dorsal fin geometric morphological features for classifying age-groups by using non-uniform B-spline. Firstly, this method uses non-uniform B-spline method to fit the dorsal fin edge curve of Irrawaddy dolphins. Then the curve is resolved to extract its derivatives, curvature and other features that describe its geometry morphology. Finally, the random forest (RF) was supplemented to classify Irrawaddy dolphins into juvenile, middle-aged and older. Experiments show that the method proposed in this paper could effectively classify older dolphins to an accuracy of 87.36%. The method proposed in this paper is non-invasive not only can help analyze the Irrawaddy dolphin's development trajectory, understand the reproductive maturity, fecundity rates, and survival rates, but also can help understand the changes of the regional ecological environment and protect the aquatic diversity.