Enhanced conditional self-attention generative adversarial network for detecting cotton plant disease in IoT-enabled crop management

Cotton plant disease identification poses challenges due to image acquisition limitations, diverse illnesses, lack of labelled data, and cost constraints. Despite these obstacles, early disease detection is crucial for farmers to enhance crop yields and mitigate financial losses. This manuscript proposes an enhanced conditional self-attention generative adversarial network (CSA-GAN) for Detecting Cotton Plant Diseases in IoT-Enabled Crop Management. Initially, IoT-generated images undergo improved bilateral textureto reduce noise and enhance image quality. Subsequently, entropy-based local binary pattern extracts relevant features for classification. The CSA-GAN model then classifies images into Normal and Diseased categories. Evaluation on a Python platform includes metrics such as accuracy, sensitivity, specificity, precision, F1-score, ROC, and computational time. Comparative analysis demonstrates that the proposed method achieves superior performance compared to existing approaches such as CNN-CPD-IoT, SVM-CPD-RBPi-IoT, and SVM-CPD-IoT, with notable improvements in accuracy and sensitivity.