<![CDATA[This study aims to develop a digital image-based system for identifying leaf spot diseases in oil palm plants using the CNN (Convolutional Neural Network) algorithm with the VGG19 architecture. The dataset consists of 330 primary oil palm leaf images categorized into three classes: leaves infected with leaf rust, healthy leaves, and leaves infected with curvularia. The dataset was divided into 64% training data, 16% validation data, and 20% testing data. The system development process includes preprocessing, data augmentation, data splitting, model training using the VGG19 architecture, and model evaluation. The training results over 200 epochs achieved an accuracy of 0.93 on the training data and 0.98 on the validation data. Model evaluation on the test data produced precision, recall, and F1-score values of 0.94, 0.81, and 0.87 for the “Leaf Rust” class; 0.84, 0.95, and 0.89 for the “Healthy Leaf” class; and 1.00 for the “Curvularia” class. The testing results indicate consistent performance, suggesting that the proposed system is effective in classifying oil palm leaf spot diseases. The developed system has the potential to be used as an early detection tool for leaf spot diseases to support the improvement of oil palm productivity.]]>
Copyrights © 2025
