A novel ensemble deep learning approach for detecting mango leaf diseases

Our innovative research addresses critical challenges in mango disease detection by developing an advanced ensemble neural network that combines EfficientNet, MobileNet, and ResNet architectures. This integrated approach overcomes the limitations of single-model systems, achieving ۹۸.۸% accuracy in identifying four mango leaf diseases: powdery mildew, anthracnose, red rust, and bacterial canker. This significantly outperforms both individual models and conventional ensemble methods. The system’s computational efficiency enables real-time disease detection on mobile devices and through IoT infrastructure, enabling farmers to implement timely interventions and optimize agrochemical applications. This technological advancement is a significant step towards sustainable mango cultivation, reducing environmental impact while improving crop yields and economic outcomes for producers.

کلیدواژه ها:

computer vision ، Deep Neural Networks ، ensemble learning ، Mango disease detection ، Precision agriculture ، transfer learning