Klasifikasi Gambar Burung Konservasi di Wilayah Papua Barat Menggunakan Transfer Learning

Abstract

Birds play an important role in maintaining the ecosystem. However, human activities such as poaching have caused some bird species to become endangered. This is rooted in the lack of public understanding of conservation bird species. The purpose of this research is to build an effective machine learning model that classifies conservation and non-conservation birds based on images, so that it is expected to improve the public's understanding of conservation and non-conservation bird species. In this study, Big Transfer (BiT), DenseNet121, and VGG16 are used as the basis for building the model. The dataset used consists of ten species that have been annotated into two classes, namely conservation and non-conservation. The model achieved the best performance with the highest accuracy obtained by the Big Transfer (BiT) model at 96.53%. The built DenseNet121 and VGG16 models have lower accuracy of 92.36% and 81.94%.