<![CDATA[Mango (Mangifera indica Linn.) is a nutrient-rich fruit, yet leaf diseases caused by microorganisms can significantly reduce crop productivity. Early detection is essential to prevent further damage and support effective disease management. This study proposes an optimized mango leaf disease prediction model using a multi-layer perceptron neural network (MLP-NN). Image-based feature extraction is performed using the Inception v3 architecture to obtain high-level color and texture features that improve classification performance. Unlike previous studies that rely solely on manually engineered features or full CNN training, this research introduces a hybrid approach that integrates deep feature extraction with MLP-NN optimization, offering a lightweight yet highly accurate alternative. Several hyperparameter combinations, including activation functions (ReLU, tanh, sigmoid) and optimization algorithms (Adam and SGD), were evaluated using the Orange platform. The optimized MLP-NN model with ReLU and Adam achieved the highest accuracy of 93.5%, demonstrating better stability and training efficiency compared to other configurations. These findings highlight the novelty and advantages of the proposed method, showing improved accuracy with lower computational cost relative to many existing approaches. This study provides an efficient solution for mango leaf disease prediction and supports future development of automated plant disease detection systems]]>
Copyrights © 2026
