Identifying Giant Clams Species using Machine Learning Techniques

Abstract

Accurate species identification is essential in preserving biodiversity. Understanding how each species can be uniquely identified determines how we can shape essential conservation efforts. One of the challenging species to identify is the Giant Clams. Due to its uniquely colored mantles and sometimes similarities in other attributes like sizes, it is challenging to distinguish each Taklobo species. A field expert is sometimes needed to identify each species correctly. The study aims to assess the possibility of automating the identification of the Giant Clams species (Taklobo) by using machine learning techniques. Different image features extraction techniques such as Scale-Invariant Feature Transform (SIFT) and Oriented FAST and Rotated Brief (ORB) were used to extract image descriptors, and color representations were used during experiments. Experimental results show that the Artificial Neural Network (ANN) with the RGB, YCbCr, HSV, CiELab color representation gained the highest accuracy rate of 89.69%.