Multimodal AI-Based Crop Health Monitoring Using Image, Weather and Soil Data for Precision Agriculture: A Review and Proposed Framework

Detecting disease symptoms in the field before they reach advanced stages gives farmers more options in how they respond. Classification of leaf images by deep learning models has already been shown to be possible. However, leaf images alone are not enough, because a leaf image does not capture the circumstances surrounding a crop’s growth. These circumstances include temperature, humidity, rainfall, moisture, soil properties and the crop’s stage of growth. This paper examines deep learning for plant disease detection, computer vision, multimodal language models, and privacy-aware and explainable AI. Using the existing literature, it proposes a framework for analysing leaf images together with structured weather, soil and crop-stage data. The framework is not complete and has not yet been tested. Its worth will have to be determined in a study that compares image-only and multimodal models on test datasets collected under actual field conditions.