Early Disease Classification of Mango Leaves Using Neural Network

Plant diseases significantly contribute to economic and productivity losses in agriculture, with 30 to 50 percent of total food production being compromised despite technological advancements. Early detection of symptoms is crucial, as untreated infections can spread, potentially affecting the entire plant. Manual observations are error-prone, necessitating the development of automated disease detection systems. Recent advancements in Deep Learning, particularly Transfer Learning using pretrained Convolutional Neural Network (CNN) architectures such as Resnet-18, Resnet-50, and SqueezeNet, have shown promising results in identifying plant leaf diseases. In our study, we propose a Transfer Learning approach using Resnet-18, Resnet-50, and SqueezeNet to identify four diseases affecting mango plant leaves-aulacaspis_tubercularis, ceroplastes_rubens, erosomiya_sp, and orthaga _ euadrusalis-along with healthy leaves. Among the architectures tested, ResNet-50, employing Stochastic Gradient Descent (SGD) as the optimizer and a learning rate of 0.001, demonstrated superior performance, achieving a test accuracy of 99.79%. This suggests the efficacy of our proposed model in accurate and efficient identification of mango plant leaf diseases, paving the way for improved agricultural management practices.