A Comprehensive Review of Crop Yield Prediction using Various Deep Learning Techniques

Abstract

Crop yield prediction gains more importance in financial evaluation at the field level to determine the strategic plans for increasing farmers’ income and import-export policies in agricultural commodities. Due to the growing concern of national food security, efficient crop yield prediction is significant in enhancing food production. With the advancement of computational technology, different methods are designed to predict crop yield, but accurate prediction still results in complex outcomes because of the dependence on complicated factors, such as pest infections, soil quality, landscape, water availability, climatic conditions, and genotype. An accurate prediction method helps farmers to select crop types, optimize resources, sustainability, and achieve market benefits, enabling them to maximize yield potential. Agricultural technology enhances the efficiency of crop yield; therefore, this survey is designed to make a thorough review analysis of the usage of different deep learning methods to predict yield by collecting 27 recent existing articles on yield prediction. The methods designed to make the prediction process are discussed along with the dataset and the performance values gained by the methods. The research methodologies are elaborated, and the research issues faced by these approaches are analyzed and discussed, which empowers scholars to develop a hybrid mechanism in the future to perform accurate yield prediction in order to increase the productivity of crops.