REVOLUTIONIZING CITRUS AGRICULTURE USING DISEASE FORECASTING THROUGH CONVOLUTIONAL NEURAL NETWORKS FOR LEAVES AND FRUITS

Abstract [English]

Citrus crops are vital contributors to the global agricultural economy. However, they are susceptible to various diseases that can significantly impact yield and quality. Early detection and management of these diseases are crucial for maintaining healthy citrus orchards. In this study, we propose a deep learning-based approach for the automated prediction of diseases affecting citrus leaves and fruits using the VGG16 convolutional neural network model. The proposed model leverages transfer learning, utilizing the pre-trained VGG16 model, which has demonstrated effectiveness in image classification tasks. We assemble a dataset comprising images of healthy citrus leavesandfruits,alongwithimagesdepicting commondiseasessuchascitruscanker,citrus greening,andcitrusblackspot.Theseimages are preprocessed and augmented to enhance model generalization and robustness. The VGG16 model is fine-tuned on the citrus dataset,wherethelastfewlayersarereplaced withcustomfullyconnectedlayersfordisease classification. During training, the model learns to extract discriminative features from citrus images, enabling it to differentiate betweenhealthyspecimensandthoseaffected by diseases. We employ data splitting techniques to ensure rigorous evaluation of themodel'sperformance,includingvalidation on separate datasets. The efficacy of the proposed model is evaluated through comprehensive experiments, including accuracy assessment, confusion matrix analysis,andcomparisonwithexisting methodologies. The results demonstrate the potential of the VGG16-based approach in accurately predicting citrus leaf and fruit diseases, thus facilitating timely intervention and management practices in citrus cultivation.