Semi-Supervised Learning in Smart Agriculture: A Systematic Literature Review

Abstract

Smart agriculture is an emerging domain that makes use of IoT to monitor the crops, creates alerts for pests, and uses enhanced ways to irrigate and increase productivity. It helps owners to analyze the fields considering multiple factors such as weather, light, and temperature. A dashboard keeps track of time for irrigation, fertilization and monitors the continuous growth of crops. The importance of machine learning in predicting agricultural KPIs in smart agriculture can hardly be underestimated. The pace of research in this domain, particularly fueled by advances in deep leaning, requires an intermittent review to present to the community. This work presents the first systematic literature review to gauge the applications of semi-supervised learning to smart agriculture. We focus on semi-supervised techniques due to their important role in labeling unlabeled data for learning tasks based on agricultural image data. We filtered 15 articles through standard SLR process and categorize the results over four semi-supervised approaches.