Monitoring Jackfruit Growth Based on YOLOv8 with Unmanned Aerial Vehicles

Jackfruit trees are usually tall, making manual monitoring methods inefficient for jackfruit cultivation. In this paper, we propose utilizing an unmanned aerial vehicle (UAV) equipped with the YOLOv8 algorithm for jackfruit growth monitoring, alongside Visual-Inertial Navigation System Fusion (Vins-Fusion) for positioning and integrated front-end and back-end trajectory optimization techniques. We collect sample data by photographing jackfruit and searching for images on the internet. The data are annotated using LabelImg and then used to train the deep learning model. The trained model is deployed on the UAV, which is equipped with a Jetson module and a depth camera, to achieve aerial detection of jackfruit ripeness and determine if they are rotten, using the YOLOv8 algorithm. Finally, we discuss the experimental results, focusing on training loss, validation loss, precision, recall, and mean average precision (mAP), and flight recognition capability. The results demonstrate that the UAV can efficiently, autonomously, and accurately detect the ripeness of jackfruit and determine whether it is rotten.