Deteksi Jenis Hama pada Daun Tanaman Mangga Menggunakan Metode You Only Look Once Versi 5

Abstract

Mango (Mangifera indica) is one of the well-known and beloved fruits, as well as an economic commodity in Indonesia. However, the cultivation process of mango plants is not always smooth and often encounters attacks from diseases and pests that can cause damage to the leaves, decrease fruit production, and even lead to the death of the plants. This has led farmers to resort to using pesticides, which can be harmful to the surrounding environment and health. Therefore, a system is needed that can utilize color and texture images of the leaves to quickly and accurately detect diseases and pests on mango leaves, allowing for better management and reduced usage and impact of pesticides.In this study, the You Only Look Once version 5 (YOLOv5) method is used to detect types of diseases and pests on mango leaves in real-time, consisting of three classes: Mango Leaf Webber, Leaf Rolling Weevil, and Javanese Grasshopper. The dataset used comprises 1,250 data points, with 1,000 data points for training and 250 data points for validation. For testing, 120 data points are used, collected through smartphone cameras. The implementation of YOLOv5 method for detecting diseases and pests on mango leaves achieved an accuracy rate of 93.3%. These results demonstrate that the system created using the YOLOv5 method performs well in detecting types of diseases and pests on mango leaves.