DefinePK

Abstract

Predicting nutrient inadequacies in crops is difficult because of the intricate relationships between soil, environment, and plant physiology. Traditional assessment approaches are inefficient and only identify problems after symptoms manifest, which results in decreased yields. This research examines the application of deep learning algorithms for emerging detection of nutrient deficiencies at both the crop and leaflet scales. At the crop scale, models utilize satellite, drone, and IoT devices for nitrogen, phosphorus, and potassium deficiency forecasting for staple crops: wheat, rice, and corn. At the leaf scale, computer vision and spectral imaging identify iron, zinc, and magnesium deficiencies prior to the manifestation of visible symptoms. This review captures the years 2015 to 2024, analysing the progress and gaps in deep learning implementations for precision agriculture along with the proposed models, strategies, and results in such advanced technologies. It presents the state-of-the-art and open problems in AI-based nutrient deficiency prediction. This research serves as a reference for stakeholders and scientists dealing with crop nutrition, management, and forecasting under the context of industrial and enterprising agriculture.