Artificial intelligence and Multispectral Imaging Technologies for Assessing Climate Impact on Maize Yield: A Literature Review

Maize production in Central America and Mexico is particularly susceptible to climate variability, especially the El Niño-Southern Oscillation (ENSO), which causes drought and irregular rainfall. This systematic literature review analyzed 111 studies, focusing on AI algorithms and multispectral imaging technologies for maize health monitoring and yield prediction. Evidence indicated that Random Forest has been repeatedly used for image segmentation and the prediction of crop and climate numerical data, often outperforming other methods. Deep learning models (ResNet50 and YOLOv3) demonstrated high accuracy (>95%) in detecting maize diseases by image recognition. Ensemble approaches, including XGBoost and hybrid models, delivered highly accurate yield predictions (R² > 0.9). However, despite these advancements, challenges persist in applying them to smallholder farms due to regional climatic and agricultural constraints. This review highlights the potential of AI-driven precision agriculture to inform which approaches can be effectively used in local adaptations.