Integrative Approaches in Modern Agriculture: Leveraging IoT and Deep Learning for Plant Disease Forecasting Under Climate Change

Climate change and severe weather events, such as droughts, frosts, and temperature extremes, have a detrimental effect on crop yields, making it harder for the agricultural sector to fulfil the world's food demands. In this regard, reliable disease outbreak forecasting is made possible by plant disease forecasting models that make use of real-time data and cutting-edge technology. These models are crucial in the development of efficient and sustainable agricultural systems. Using DL classifiers, class imbalance handling strategies, and preprocessing methods, this work investigates a full pipeline for plant disease diagnosis. Using K-means segmentation, everyone were able to extract image features, calculate average intensities of rgb and hue, and describe the texture of hue lesion using ASM, Entropy, Dissimilarity, Homogeneity, and Energy. By including image enhancing filters and adding more layers to a CNN-based B0-EfficientNet architecture, a specialised classification model was created. Classification accuracy was 99.36%, and the suggested model outperformed competing CNN models in terms of accuracy, recall, precision, and AUC. These results demonstrate that B0-EfficientNet is a powerful tool for predicting plant diseases and stress the need for AI-powered responses to lessen the impact of climate change on crop yields.