Modern agriculture faces critical challenges in meeting global food demands while maintaining environmental sustainability. This study presents an AI- driven framework integrating computer vision and IoT technologies to optimize crop management. Our system employs multimodal sensor networks capturing real-time soil moisture (accuracy: ±1.8%), nutrient levels, and crop health data, combined with drone-based hyperspectral imaging for pest detection (F1- score: 0.92). A novel deep learning architecture processes this data through spatial-temporal graph convolutional networks, achieving 94.3% accuracy in yield prediction across three major crops. The IoT infrastructure reduces water usage by 37% through adaptive irrigation control, while computer vision modules decrease pesticide application by 42% via targeted treatment zones. Comparative trials demonstrate 28% higher resource efficiency than conventional precision agriculture methods [1]. This approach addresses key limitations in current systems by enabling sub-field-level management decisions through edge computing devices with 89% lower latency than cloud-based alternatives. The framework’s scalability and interoperability with existing farm equipment position it as a viable solution for sustainable intensification of agricultural production.