Deep Learning-Based Plant Disease Detection and Prediction Model for Smart Agricultural Applications

Abstract

Diseases of plants are a significant problem that is impacting the production of agriculture, the quality of food and the sustainability of an economy worldwide. Disease diagnosis with accurate and quick diagnosis is necessary to reduce the losses of crops and be supportive to precision agricultural practices. This work introduces a deep learning-based plant disease detection and prediction system to be used in smart farming. The following framework combines image processing, data enhancement, and classification using a convolutional neural network (CNN) to monitor the health of crops intelligently and predict diseases. Agricultural leaf image datasets that were publicly available with several categories of crops and diseases were used to carry out experimental evaluation. Image resizing, normalization, and noise filtering as preprocessing steps were used to increase image quality, and augmentation methods were used to increase the diversity of datasets and decrease overfitting. The accuracy, precision, recall, F1-score, and ROC-AUC measures were used in the evaluation of the proposed CNN architecture. Experiments showed better performance than traditional machine learning methods like the Support Vector Machine (SVM), Random Forest (RF) and K-Nearest Neighbor (KNN). The proposed model had a classification accuracy of 98.1 percent, precision of 97.6 percent, recall of 97.3 percent and F1-score of 97.4 percent. The power and the ability to generalize the framework was verified through cross-validation analysis. The intelligent system developed can be used to aid the self-monitoring of crops, prediction of diseases early, targeted farming, and sustainable management of agriculture. The suggested solution is a step towards AI-assisted smart agriculture systems that can enhance crop yield and minimize losses in agriculture.