Palm Fruit Ripeness Detection and Counting Using YOLOv8 Algorithm in PTPN IV Medan North Sumatera Indonesia

Oil palm plantations have become a crucial sector in the Indonesian economy. This research investigates the effectiveness of YOLOv8 in identifying the ripeness of oil palm fruits. The goal is to create a system that uses the YOLOv8 Deep Learning algorithm to detect the ripeness level of oil palm fruits. An automatic detection system will simplify and accelerate the sorting or classifying of oil palm fruits. This research utilizes datasets sourced from the oil palm plantation repository of PTPN IV Medan, North Sumatera, Indonesia. The dataset collected for this research consists of 85 images with a size of 640x640. The dataset is loaded into Roboflow and undergoes pre-processing and augmentation. The total dataset obtained after augmentation is 185 images. After obtaining the API from Roboflow and integrating it into Google Colab for YOLOv8 training, testing was performed on the provided images. The results indicated that YOLOv8 achieved an accuracy of 0.991 (99.1%) in detecting ripe, unripe, and rotten oil palm fruit. The high value of accuracy leads to the potential to increase the efficiency of oil palm plantation management and become the basis for developing AI-based agricultural technology. With further updates, the YOLOv8 model could become a handy tool in oil palm plantation monitoring and management in the future.