Research Area:  Machine Learning

Abstract:

Considering the variability of the farming resources such as soil, fertilizer and weather conditions including crops. Proper utilization of these resources for high yield is paramount. Since practices based on human experience lead to low crop yield, especially the inconsistent climate condition. Though, the advent of smart farming and precision agriculture driven by machine learning provides an avenue for automated solutions to this problem. Therefore, this article provides a comprehensive survey of the existing smart farming models for precision agriculture. The survey focused on machine learning approaches such as supervised, unsupervised, deep learning and ensemble techniques employed in smart farming. In addition, smart farming innovations for predicting key agrarian practices such as soil, crop yield, and fertilizer including weather and irrigation models were reviewed. Thus, poor performance of some models is observed and attributed to the dataset used, negligence of pre-processing and feature extraction stages by some researchers. Consequently, the survey projected a machine learning procedure that if duly followed would prevent such limitations in smart farming models. This survey has demonstrated how machine learning can automate agricultural practices and enhance crop quantity and quality with minimal human labour. It finally outlined the challenges and prospects of smart farming models for researchers to explore.

Keywords:  
Smart-farming
Precision-agriculture
Ensemble-learning
Crop-soil prediction
Fertilizer-recommendation
Climatic-conditions
Machine Learning
Deep Learning

Author(s) Name:  Dekera Kenneth Kwaghtyo & Christopher Ifeanyi Eke

Publisher name:  Springer

DOI:  10.1007/s10462-022-10266-6

Volume Information:  

Paper Link:   https://link.springer.com/article/10.1007/s10462-022-10266-6