<![CDATA[Papaya leaf diseases, such as Papaya Ringspot Virus and Begomovirus, have a significant impact on papaya production and quality. Manual identification is often ineffective due to the visual similarity of symptoms. This study proposes a system for classifying papaya leaf diseases using a Convolutional Neural Network (CNN) with an ensemble learning method that aims to improve accuracy and provide accurate and stable predictions on new data that has not been seen during training. Three CNN architectures, namely VGG16, ResNet-34, and MobileNetV2, were trained individually with hyperparameter optimization using grid search. After training the individual models, the three models were combined (ensemble) using the soft voting method. The results showed that the best individual model was MobileNetV2, achieving 98% on all accuracy, precision, recall, and f1-score metrics, but the model with the most optimal performance was achieved by the ensemble model. The ensemble model achieved 99% accuracy, 98% precision, 99% recall, and 99% f1-score, these results demonstrate a significant improvement over the three individual trained models and outperform the ResNet-152v2 architecture reported in previous research]]>
Copyrights © 2026
