Establishing a Standard Operating Procedure (SOP) for Palm Oil Plantation FFB Image Capture: Utilizing YOLOv8 for Counting and Ripeness Classification

Almost all oil palm harvest management is still done manually today, especially in the monitoring stage where oil palm farmers assess the ripeness of oil palm fresh fruit bunches (FFB) in the plantation. The monitoring stage, which is a crucial step, requires re-checking the ripeness of fresh fruit bunches (FFB) at the collection point (TPH) to ensure that the bunches brought to the factory are ripe. This research focuses on creating an SOP to be implemented in the plantation so that the AI in the automated system, also created in this research, can be used to optimally detect, calculate, and classify the ripeness of the FFB. By using the YOLO8 algorithm based on deep learning and CNN, the dataset used consists of 4257 images taken directly from PTPN IV plantations. The results of this study showed good model performance, with an accuracy of 96.56%. The application of this model is expected to increase the efficiency and accuracy of FFB quality monitoring, reduce losses, and support better data-based decision-making.