Leveraging Generative AI for Accurate Crop Disease Detection

Crop diseases are one of the major threats to agriculture. They severely impact productivity and sustainability. This paper presents a Artificial Intelligence-driven framework for accurate crop disease detection using Conditional Generative Adversarial Networks (cGANs) and ResNet18. cGANs are used to generate high-quality synthetic images of diseased crops, which enrich datasets usually constrained by data scarcity. These enriched data are utilized in fine-tuning ResNet18. Notable classification accuracies are observed for both training, at 96.15%, and validation set, at 95.26%, despite having changed environmental conditions. Further integrating with the Gemini Application Program Interface, it is designed to deliver real-time and actionable disease-related information. It comes equipped with a friendly web interface developed in Streamlit to provide farmers easy access for the disease diagnosis and sustainable remedies. It is an easy-to-scale solution that exhibits how Generative AI can change agriculture through timely interventions and the development of healthy crops with sustainable agriculture practices.