Modern Computer Vision for Oil Palm Tree Health Surveillance using YOLOv5

Oil palm (Elaeis guineensis) is an important species of palm vegetation for bio-energy agribusiness. Although oil palm tree plantations have expanded rapidly, especially in tropical countries to meet the need for biofuels, issues like diseases can have a negative impact on the industry by reducing productivity and survival rates of palm trees. Therefore, a regular tree counting is needed for inventory and health monitoring. The rapid advancement of deep learning-based computer vision and remote sensing technology has made it possible to automate tree counting. In this paper, we use YOLOv5 model for counting oil palm trees from Papua, Indonesia. The image data are divided into five classes, namely healthy, smallish, yellowish, mismanaged, and dead palms. We achieve average Fl-score of 0.895, which outperformed Faster R-CNN (0.706) and CNN ResNet-101 (0.493). The strength of YOLOv5 model is high precision for all the five classes, which is above 0.961. This application provides fast, robust, and accurate oil palm tree counting that can be applied elsewhere in the world.