Leveraging IoT & CNN-CBAM for Early Blight Detection and Optimal Fungicide Application in Potato and Tomato Crops

This study focuses on the problem of early blight detection in potato and tomato crops, which is a worldwide issue resulting in substantial agricultural losses. Conventional techniques such as Support Vector Machine (SVM) and Random Forest Classifier (RFC) have constraints in accurately identifying early blight, which can lead to inadequate use of fungicides. In order to address these restrictions, this study suggests a structure that combines the Internet of Things (IoT) with a Convolutional Neural Network (CNN) that is improved by using a Convolutional Block Attention Module (CBAM). The CNN-CBAM model, trained on Kaggle datasets, attained a precision of 0.95, an F1 score of 0.93, and a recall of 0.94, surpassing the accuracy rates of the SVM and RFC models, which were 0.87 and 0.89, respectively. The suggested method improves the early identification of early blight and the use of fungicides, hence enhancing crop management. It provides enhanced precision and dependability compared to current techniques, rendering it a useful asset for contemporary agriculture and a basis for future investigations.