<![CDATA[Rice is a strategic commodity in Indonesia, yet its productivity is often threatened by leaf diseases such as Bacterial Leaf Blight, Brown Spot, Narrow Brown Spot, and Tungro. Conventional identification conducted by farmers is subjective and may delay proper treatment. This study implements transfer learning using the MobileNetV2 architecture for rice leaf disease classification. The dataset was collected directly from rice fields in OKU Timur, South Sumatra, consisting of five classes (Bacterial Leaf Blight, Brown Spot, Healthy, Narrow Brown Spot, and Tungro), each containing 300 images except Narrow Brown Spot, which was balanced through light augmentation. All images underwent cropping, resizing to 224×224 pixels, and normalization before being split into training, validation, and testing sets. The proposed model achieved 99.11% validation accuracy and 100% testing accuracy, with near-perfect precision, recall, and f1-score. The model was then deployed into a user-friendly web application using Streamlit, enabling farmers to upload rice leaf images for instant classification and recommended treatments. These findings demonstrate that MobileNetV2 with transfer learning provides accurate early detection of rice leaf diseases and supports better decision-making in rice cultivation.]]>
Copyrights © 2025
