A novel age-group classification method for Irrawaddy dolphins based on dorsal fin shape features

Abstract

The Irrawaddy dolphin is an endangered marine mammal species; therefore, there is an urgent need to take protective measures, especially in terms of population breeding and evolution. To address this, it is important to understand the age group structure of populations. Unlike biological individual identification and biological object detection based on pattern classification methods, a new age-group classification (AGC) method was developed to classify Irrawaddy dolphins into three age groups: older, middle-aged, and juvenile. Taking into account the relation between the dorsal fin shape features of Irrawaddy dolphins and their age, the AGC method constructed several dorsal fin geometric morphological features, such as leading edge length and dorsal fin height, using edge extraction and curve fitting of dolphin images. After performing a multicollinearity test on these features, nine effective features were obtained. A model was then trained to classify Irrawaddy dolphins according to their age groups. The experimental results demonstrated that the AGC method has a high classification accuracy of 80.20% for older dolphins. In contrast to individual identification and object detection methods, the proposed AGC method facilitates the analysis of population structure stability and dynamics by classifying Irrawaddy dolphins by age.