| dc.description.abstract | Oil palm (Elaeis guineensis jacq) is one of the most important plantation crops in Indonesia, because it produces vegetable oil which is a daily necessity for Indonesian people and has made a significant contribution to the Indonesian economy in recent years. The growth of oil palm plants is still disturbed by the emergence of pest and disease attacks on the leaves of oil palm plants. Yellow patch disease caused by the pathogen Culvularia Eragrostidis, Drechslera Halotes is discussed in this study. However, detection of oil palm leaf diseases is still done manually by observing with the human eye based on the characteristics and changes in the color of the oil palm leaf spots. The manual recognition process is time consuming and prone to inaccuracies or errors due to limited human resources and human judgment of different colors. Digital image processing systems are needed to support the detection process which must be done quickly and accurately. In this study, the authors used the SSD-MobileNet V2 method to detect the level of damage to oil palm plant disease types from their leaf images in real-time. Detection is carried out on three levels of yellow patch disease, namely mild yellow patch, severe yellow patch, and healthy leaves. Leaf images were tested at different positions for each disease type using a camera at a distance of 10 cm or more. Based on the tests that have been carried out, the results show that the system built using the SSD-MobileNet method is able to detect images of yellow patch disease types with an accuracy of 95.3%. | en_US |
