Drone-Based Monitoring of Palm Fruits Ripeness using YOLOv8

Proper and timely ripeness assessment of palm fruits is essential to ensure optimum oil production, minimise post-harvest losses and achieve sustainable palm cultivation. The traditional ripeness evaluation method mainly relies on manual inspection, which is time-consuming, labour-intensive, and prone to human error. This study proposes an automatic and improved palm fruit ripeness classification using a novel drone-based monitoring framework that combines the LiDAR (Light Detection and Ranging) 3D mapping technology with deep learning-based object detection technology that employs a convolutional neural network (CNN) architecture known as You Only Look Once (YOLO). Aerial imagery and dense 3D point cloud data were gathered using unmanned aerial vehicles (UAVs) fitted with LiDAR sensors and high-resolution RGB cameras over oil palm plantations. The data was then processed using a YOLO-based detection pipeline trained on a curated dataset of 4200 annotated palm fruit bunches across three ripeness classes: unripe, ripe, and overripe. Image-based colour and texture features were also incorporated with LiDAR-derived structural features (bunch height, canopy density, spatial distribution) to improve classification accuracy.