Artificial intelligence for sustainable agriculture: A review of machine learning and deep learning techniques for crop growth and yield prediction

Abstract

The climate change, the lack of resources, soil decline, the population growth, and food security have put additional pressure on agriculture, which poses an urgent necessity to improve the accuracy and sustainability of systems of crop growth and crop yield projections. Machine Learning and Deep Learning are types of Artificial Intelligence that have become a groundbreaking method to overcome these issues in Sustainable Agriculture and Precision Agriculture. It is a PRISMA-based literature review of recent advances in Crop Growth Prediction and Crop Yield Prediction based on data-driven farming strategies with the emphasis on the research published in the age of Smart Farming, Digital Agriculture, and Climate-Smart Agriculture. The literature review was systematic in terms of tackling literature connected to machine learning, deep learning, remote sensing, UAV imagery, satellite imagery, Internet of Things sensors, and Agricultural Big Data to predictive modeling in agriculture. Typically used algorithms are Random Forest, Support Vector Machine, Gradient Boosting, Ensemble Learning, Convolutional Neural Network, Long Short-Term Memory, Reinforcement Learning, Transfer Learning, and Vision Transformer architectures. The results mean that Deep Learning-based architecture, in general, and hybrid CNN-LSTM and multimodal architectures in particular, are typically more successful in Crop Yield Prediction than classical statistical approaches, especially with the inclusion of weather, soil, vegetation index, and remote sensing data.