Automatic leaf diseases detection and classification of cucumber leaves using internet of things and machine learning models

Automation of agriculture with the use of cutting-edge technology is a growing research area. It addresses the issue of better yields and tries to mitigate the negative impact due to climatic changes, attacks of diseases, and pests in crops. Hence to overcome the problem of disease attacks, this research proposes an automatic leaf disease detection and classification system using a web app that can help the farmer to identify the occurrence of leaf diseases remotely. Further performance matrices like confusion matrix, overall classification accuracy, precision, sensitivity, specificity, and ROC-AUC score have been calculated to test the efficacy of models. The simulated results proved that the CNN with 10-fold cross-validation has got an accuracy of 99.47% and it significantly outperforms other existing counterparts. The data collected from both environments have been compared and analysed. This study offers a real-time application of the internet of things and machine learning in agriculture.