<![CDATA[This study investigates the application of Artificial Intelligence, specifically Convolutional Neural Networks (CNN), to support early detection of shallot leaf diseases, namely Moler and Purple Spot, which are commonly identified through manual visual inspection and are prone to subjectivity. The MobileNetV2 architecture is employed using a transfer learning approach on a publicly available shallot leaf image dataset. The research stages include data preprocessing, image augmentation, model training with a fine-tuning strategy, and implementation within a web-based system. Experimental results on the test dataset indicate that the proposed model achieved an accuracy of 99.07%. In particular, the model demonstrated high recall in detecting Moler disease and high precision in identifying Purple Spot disease. These findings suggest that lightweight architectures such as MobileNetV2 are suitable for efficient and accurate plant disease detection with relatively low computational requirements.]]>
Copyrights © 2026
