Publication

Abstract : Coconut trees play a critical role in tropical areas, serving as sources of food, shelter, and economic means. Diseases in coconut trees, though, impact the yield of crops remarkably, and therefore early and precise identification is significant. This work explores the performance of deep learning models-MobileNetV2, InceptionV3, and VGG 19-in classifying coconut tree diseases. This work employs advanced deep learning techniques, including data augmentation and L2 regularization, to enhance disease management strategies in coconut plantations. This developed model has been examined for the five major diseases: Bud Root Dropping, Bud Rot, Gray Leaf Spot, Leaf Rot, and Stem Bleeding, by deploying precision, recall, and F1-score for evaluating the model. We also experimented with four different activation functions-ReLU, LeakyReLU, Swish, and ELU. Our best model has achieved an accuracy of 99.88% in classifying the coconut tree diseases, demonstrating great performance.