Klasifikasi Spesies Burung Endemik Menggunakan Metode Retinanet

Abstract

Endemic birds are one of the animals that are quite popular with the people of Indonesia. There are hundreds of species of endemic birds in Indonesia that have characteristics, shapes, and colors that are similar to one another, so it is sometimes difficult to distinguish endemic bird species directly. In this study, there were three species of birds to be classified, namely the blue fronted redstart, the common redstart, and the taiwan vivid niltava where the three birds belong to the Muscicapidae family or more commonly known as the tledekan bird and the three birds have color and shape characteristics that are almost similar to each other so that they are not mistaken in cultivating these birds. Therefore, an information technology system is needed that uses digital image processing to properly identify endemic bird species. The classification method used in this study is RetinaNet. This study uses a total of 600 data which is divided into training data and testing data, of which 480 data are used as training data and 120 data are used as testing data. After testing in this study, it can be concluded that the results of this study can be considered quite good in classifying the three endemic bird species studied and show that the method used is capable of classifying endemic birds, namely with an accuracy of 93.3%.