Implementasi Algoritma Yolo untuk Deteksi Buah Durian dan Manggis

Abstrak

This study aims to implement the YOLOv8 algorithm in detecting images of durian and mangosteen fruits. The research methodology includes literature review, image collection, data processing, YOLOv8 algorithm implementation, model evaluation on validation data, and drawing conclusions. Image collection is done through online sources, and data is processed through annotation, pre-processing, and augmentation using the Roboflow platform before exporting to YOLOv8 format. The algorithm implementation is carried out in Google Colab with model training, object detection, and evaluation stages on validation data. Evaluation results include accuracy, recall, precision, and F1 score values, with model performance evaluated using mean average precision (mAP) metric. The results indicate that the model can recognize objects well, with a mAP above 0.27%. This study successfully implements YOLOv8 for durian and mangosteen fruit detection with satisfactory evaluation results.