An integrated AI framework for growth stage-aware corn leaf disease identification and environmental impact prediction

Abstract Corn's is one of the world's most important cereal crops, yet accurate identification of leaf diseases and stress symptoms under field conditions remains challenging because of complex backgrounds, variable illumination, and overlapping visual characteristics. This study proposes an integrated artificial intelligence framework for corn leaf condition identification across different growth stages. First, a new Corn Leaf Disease Forms (CLDF) dataset containing 2,903 field images was developed, representing four leaf condition classes: Healthy (HT), Common Rust (CR), Nitrogen Deficiency (ND), and Spodoptera frugiperda (SF) damage. To improve image quality, an advanced image processing enhancement framework (AIPEF) was introduced for background removal, noise reduction, and image enhancement. An enhanced P-YOLOv11s-MD-SiLU model, incorporating MobileNetV3 and a modified dynamic SiLU activation function, was then developed for corn leaf condition detection. The proposed model achieved a Map0.5 of 94.90%, outperforming the baseline models while reducing computational cost. For practical deployment, the trained model was integrated into a mobile augmented reality application for real-time field diagnosis. Furthermore, future leaf condition occurrence was forecast using an ARIMA time-series model based on historical environmental observations, indicating increased risks of spodoptera frugiperda damage during 2028–2030, nitrogen deficiency during 2027 and 2030, and the highest common rust occurrence in 2026, followed by a gradual decline through 2030. The proposed framework provides an effective tool for automated corn health monitoring and supports precision crop management and agricultural decision-making.