| dc.description.abstract | The wide variety of lovebird species in Indonesia often makes it difficult for the
general public to distinguish between them, highlighting the need for a system capable
of classifying lovebird types. Therefore, this study aims to classify lovebird species in
digital image form using the YOLOv8 (You Only Look Once version 8) algorithm and
implement the model into an Android-based application. Eight lovebird species are
classified: Lutino, Blue Personata, Biola Green, Biola Blue, Euwing Blue, Euwing
Green, ParBlue, and ParBlue Euwing. Initially, each species had 100 images, which
were then augmented to increase the dataset to approximately 600 images per class.
The dataset was labeled according to the bird species and divided into three parts:
80% for training, 10% for validation, and 10% for testing. The model was trained for
50 epochs using the training dataset, and the resulting model in PyTorch (.pt) format
was converted to TensorFlow Lite format to enable integration into an Android
application. The test results show that the model is capable of classifying lovebird
species with high accuracy, achieving a mAP@0.5 score of 0.992, or 99.2% accuracy.
The decreasing values of box loss, classification loss, and distribution focal loss
throughout the training process indicate stable and effective model learning. The
developed Android application is also capable of performing real-time detection,
although it still has limitations in detecting multiple objects simultaneously due to the
birds' continuous movement. Overall, the system is considered feasible as an
automatic tool for identifying lovebird species. | en_US |
