An AI-Powered Crop Health Monitoring System Using NDVI and Predictive Soil-Climate Modeling

Agriculture is under pressure to increase yield while facing climate variability and limited resources. Precision agriculture is seen as a solution to balance productivity with sustainability. AI, IoT and spectral imaging play a key role in modernizing agriculture. These technologies provide real-time data and smart decision-making tools for farmers. Traditional farming methods often detect crop stress too late. They also waste water through inefficient irrigation, leading to yield loss. NDVI imaging, soil-climate sensors and machine learning offer new possibilities. Together they enable cost effective solutions suitable for smallholder farmers. This research proposes a lowcost hybrid monitoring and irrigation system. It is built on a Raspberry Pi and can run offline, making it accessible in rural areas. The framework combines NDVI imagining, soil-climate sensing and AI models. CNN is used for disease detection, while KNN supports irrigation scheduling. Experiments achieved 94.2% CNN training accuracy and 91.6% validation accuracy. This confirms strong performance in detecting plant diseases under varied conditions. NDVI values between 0.58 and 0.71 clearly separated healthy from stressed plants. This threshold helps identify vegetation stress at an early stage. KNN based irrigation prediction reached 98.4% accuracy. An economic feasibility analysis shows a payback period of 18-24 months for a 0.5-acre farmer. The proposed NDVI thresholding and privacy-preserving model updates enhance practical deployment and real-world scalability.