<![CDATA[Tomato (Solanum lycopersicum) is highly vulnerable to a range of foliar diseases that can reduce yield and hinder crop quality, particularly when early symptoms are difficult to distinguish in field conditions. To address this challenge, this study develops a predictive model for automatic tomato leaf disease classification using a lightweight EfficientNetB0 architecture. The dataset consists of 5,967 images from nine categories, combining 70% publicly available Kaggle PlantVillage data and 30% real-field images captured under natural outdoor illumination. The methodological pipeline includes preprocessing, data augmentation, and transfer learning, followed by fine-tuning of the upper layers of EfficientNetB0 to improve its ability to generalize toward field-specific variations such as uneven lighting and complex backgrounds. Evaluation results show that the model achieves an accuracy of 89%, with macro-average and weighted-average scores of 90%. These findings demonstrate that EfficientNetB0 provides an effective balance between predictive accuracy and computational efficiency, supporting its potential deployment in early detection systems and edge-based agricultural applications for real-time tomato disease monitoring.]]>
Copyrights © 2025
