PlantHealthAI: An Integrated System for Plant Disease Detection, Severity Prediction with Knowledgebase Chatbot Support

Plant Disease detection and severity estimation is crucial for mitigating the impact on crop yield. This research focuses on the development of a deep learning-based model to detect the incidence level of early leaf spot disease in groundnut crops. A real-field dataset is created for evaluating our methodology by inducing the early leaf spot disease to susceptible plants in a controlled and observatory environment. The developed model achieved a precision and recall rates of 98 % and 95 %, respectively. Additionally, the developed severity estimation model with the custom metric demonstrated promising results, with a maximum absolute difference of 0.5 between the ground truth severity measure and the predicted severity measure. Furthermore, a user-based query feedback system backed by a comprehensive knowledge base was integrated, enabling users to acquire information on various aspects of the detected disease. This feature enhances the usability and accessibility of our solution, empowering farmers to make informed decisions for effective disease management and unlocks new potential in promoting sustainable crop protection practices and thereby positively impacting the agricultural sector.