Automated System for Detecting, Identifying, and Preventing Cotton Leaf and Boll Diseases Using Deep Learning

The Internet of Things (IoT) technology facilitates real-time data collection through sensors, enhancing disease detection and management accuracy and crop efficiency. Pest infestations in cotton crops adversely impact its production and ultimately affect the nation's economy which depends on agriculture. Previous research has primarily focused on detecting cotton leaf diseases through imagery, while this study addresses both cotton leaf and its boll diseases. An existing dataset of cotton leaf images was customized by incorporating classes of cotton boll images, resulting in a comprehensive dataset of 7289 images categorized into eight distinct classes, including healthy and diseased leaves and bolls. Deep learning models, including VGG16, InceptionV3, and a tailored model, were applied to evaluate the accuracy of disease identification and the detection of infection levels. The tailored model demonstrated performance comparable to pre-trained models, achieving high accuracy in classification tasks. The IoT-integrated, perceptionbased decision system proposed in this study enables early detection of diseases and supports the implementation of preventive measures, contributing to improved crop yields and mitigating the economic losses caused by pest infestations and diseases.