Artificial Intelligence for Crop Yield Estimation: Models, Challenges, and Future Directions

Accurate yield forecasting is a top priority for modern farmers, agribusiness specialists and agricultural regulators. We will be able to manage farms more simply due to increased access to massive amounts of data, improved machine learning (ML) and deep learning (DL) techniques, and a wealth of information for crop yield prediction. The classification of a method as an ensemble, hybrid, or conventional machine learning approach depends on the fundamental algorithm used. This research examines the concept of explainable AI (XAI) and notable advancements in deep learning frameworks and architectures, including convolutional, recurrent, and attention-based neural networks. The article discusses key components and methods for evaluating the capabilities of machine learning and deep learning systems for data collection, modelling and prediction. The review highlights several common issues with yield prediction, such as insufficient data, an inability to quantify model uncertainty or comprehend or model results, an inability to apply results to other crops, and an inability to transition from research to commercial operational contexts.