| dc.description.abstract | Guava (Psidium guajava) is a popular fruit in tropical and subtropical regions due to its sweet flavour. Despite its good market potential, intensive planting of guava in Indonesia is still rare, resulting in low production. One of the main obstacles in increasing productivity is disease infestation, which is usually diagnosed based on morphological symptoms, but is often inaccurate if only using naked eye observation. Therefore, a system is needed that can detect diseases on guava leaves and fruits more precisely. This research implements the You Only Look Once (YOLO) algorithm version 8 to detect four types of diseases on guava plants: anthracnose, phytophthora (water mould), red rust, and scab (fruit spot) with a total of 1,920 data used consisting of 1,536 data as train data, 192 data as validation data, 192. The results show that the YOLOv8 algorithm is able to detect diseases in real-time with an accuracy value of 94.27%, precision 94.60%, recall 94.26%, and F1-Score 94.31%. This model is also able to predict more than one disease object in one frame provided that the object is within the optimal distance in front of the camera. These results show that the system created using the YOLO V-8 algorithm has succeeded well in detecting diseases in guava plants. | en_US |
