Review of machine learning and deep learning models in agriculture

Abstract

Machine learning (ML) refers to the processes that enable computers to think based on various learning methods. It can be also called domain which is a subset of Artificial Intelligence (AI). Deep learning (DL) has been a promising, new and modern technique for data analysis in recent years. It can be shown as the improved version of Artificial Neural Networks (ANN) which is one of the popular AI methods of today. The population of the world is increasing day by day and the importance of agriculture is also increasing in parallel. Because of this, many researchers have focused on this issue and have tried to apply machine learning and deep learning methods in agriculture under the name of smart farm technologies both to increase agricultural production and to solve some challenges of agriculture. In this study, it is aimed to give detailed information about these up-to-date studies. 77 articles based on machine learning and deep learning algorithms in the agriculture field and published in IEEE Xplore, ScienceDirect, Web of Science and Scopus publication databases between 2016 and 2020 years were reviewed. The articles were classified under five categories as plant recognition, disease detection, weed and pest detection, soil mapping-drought index, and yield forecast. They were examined in detail in terms of machine learning/deep learning architectures, data sets, performance metrics (Accuracy, Precision, Recall, F-Score, R2, MAPE, RMSE, MAE), and the obtained experimental results. Based on the examined articles, the most popular methods, used data sets/types, chosen performance criteria, and performance results among the existing studies are presented. It is seen that the number of AI-based applications related to agriculture is increasing compared to the past and the sustainability in productivity is so promising.