Reliable and timely crop-yield prediction and crop mapping are crucial for food security and decision making in the food industry and in agro-environmental management. The global coverage, rich spectral and spatial information and repetitive nature of remote sensing (RS) data have made them effective tools for mapping crop extent and predicting yield before harvesting. Advanced machine-learning methods, particularly deep learning (DL), can accurately represent the complex features essential for crop mapping and yield predictions by accounting for the nonlinear relationships between variables. The DL algorithm has attained remarkable success in different fields of RS and its use in crop monitoring is also increasing. Although a few reviews cover the use of DL techniques in broader RS and agricultural applications, only a small number of references are made to RS-based crop-mapping and yield-prediction studies. A few recently conducted reviews attempted to provide overviews of the applications of DL in crop-yield prediction. However, they did not cover crop mapping and did not consider some of the critical attributes that reveal the essential issues in the field. This study is one of the first in the literature to provide a thorough systematic review of the important scientific works related to state-of-the-art DL techniques and RS in crop mapping and yield estimation. This review systematically identified 90 papers from databases of peer-reviewed scientific publications and comprehensively reviewed the aspects related to the employed platforms, sensors, input features, architectures, frameworks, training data, spatial distributions of study sites, output scales, evaluation metrics and performances. The review suggests that multiple DL-based solutions using different RS data and DL architectures have been developed in recent years, thereby providing reliable solutions for crop mapping and yield prediction. However, challenges related to scarce training data, the development of effective, efficient and generalisable models and the transparency of predictions should be addressed to implement these solutions at scale for diverse locations and crops.

The timely and reliable prediction of crop yields on a larger scale is crucial for ensuring a stable food supply and food security. In the last few years, many studies have demonstrated that deep learning can offer reliable solutions for crop yield prediction. However, a key challenge in applying deep-learning models to crop yield prediction is their reliance on extensive training data, which are often lacking in many parts of the world. To address this challenge, this study introduces TrAdaBoost.R2, along with fine-tuning and domain-adversarial neural network deep-transfer-learning strategies, for predicting the winter wheat yield across diverse climatic zones in the USA. All methods used the bidirectional LSTM (BiLSTM) architecture to leverage its sequential feature extraction capabilities. The proposed transfer-learning approaches outperformed the baseline deep-learning model, with mean absolute error reductions ranging from 9% to 28%, demonstrating the effectiveness of these methods. Furthermore, the results demonstrate that the semi-supervised transfer-learning approach using the two-stage version of TrAdaBoost.R2 and fine-tuning achieved a superior performance compared to the domain-adversarial neural network and standard TrAdaBoost.R2. Additionally, the study offers insights for improving the accuracy and generalizability of crop yield prediction models in diverse agricultural landscapes across different regions.

Verticillium wilt, caused by Verticillium dahliae , poses a major threat to cotton production worldwide. This study presents a framework for predicting Verticillium wilt incidence across multiple cotton fields by combining Sentinel‐2 imagery, preseason inoculum, and weather variables with machine learning models. Three machine‐learning models: random forest, support vector machine, and extreme gradient boosting were evaluated using data from seven commercial cotton fields in northern New South Wales, Australia. Model performance was assessed under two validation schemes: fivefold cross‐validation to assess within‐field interpolation performance and leave‐one‐field‐out validation to evaluate extrapolation across fields. Random forest achieved the highest accuracy (mean overall accuracy = 0.63, mean weighted F1 = 0.64 for interpolation; mean overall accuracy = 0.58, mean weighted F1 = 0.56 for extrapolation) when all data sources were combined. Model interpretability was assessed using Shapley additive explanations, which identified preseason inoculum, near‐infrared and red‐edge bands are the most influential predictors. Results highlight the advantage of integrating multisource data for disease prediction and the importance of rigorous validation to avoid overestimation of model performance—something that is commonly overlooked in published disease studies. The approach offers a scalable, nondestructive framework for operational monitoring of Verticillium wilt in cotton.

The red-edge (RE) spectral region has become a central component of agricultural remote sensing because it captures physiologically meaningful changes in chlorophyll content, canopy structure and vegetation functioning. The availability of dedicated RE bands on modern multispectral satellites and advances in hyperspectral sensing have stimulated widespread applications for crop monitoring, nutrient assessment, stress detection and yield prediction. However, reported improvements over conventional visible–near-infrared (VIS–NIR) approaches remain highly variable, and the mechanisms governing when and why RE information provides additional value are often poorly synthesised. This review presents a conceptual framework that links the physical and physiological basis of RE reflectance with its condition-dependent agronomic performance and its emerging role within modern machine learning (ML) systems. We first examine how pigment absorption, canopy structure and sensor characteristics jointly determine the representation of RE information from hyperspectral measurements to operational multispectral observations. We then synthesise evidence demonstrating that the agronomic value of RE information is strongly dependent on crop characteristics, phenological stage, environmental conditions and observation geometry, explaining much of the variability reported across previous studies. Finally, we show how recent advances in ML and explainable artificial intelligence have changed the interpretation of RE information. Rather than evaluating RE-derived vegetation indices in isolation, contemporary predictive frameworks integrate RE observations with complementary spectral, climatic, structural and temporal predictors, allowing their physiological contribution to be quantified within multidimensional models. We conclude that future value of RE remote sensing will require not only continued advances in spectral measurement and vegetation index development, but also improved interpretation, transferability and operational integration of physiologically meaningful RE information within explainable, multi-source agricultural monitoring systems.