S Leekar1,2, TA Jensen1, A Chumpia1,3, S Khawprateep2 and BL Schroeder1
1University of Southern Queensland, Toowoomba, Qld 4350; Somwang.Leekar@unisq.edu.au
2Faculty of Agriculture at Kamphaeng Sean, Kasetsart University, Thailand
3Faculty of Engineering at Kamphaeng Sean, Kasetsart University, Thailand Even though sugarcane is a major agricultural crop in Thailand, Thai sugarcane growers have recently endured high production costs with reduced income. For this reason, there have been many studies to assist farmers to achieve greater efficiencies on-farm. Recently, unmanned aerial vehicle (UAV) technology and machine learning (ML) have been introduced into agriculture. ML is a powerful method that can help researchers analyse large spatial datasets including those collected by purpose-built agricultural UAVs. This review was undertaken to assess the potential for the use of ML in sugarcane production in Thailand by using various keywords in literature and scholastic databases. These included ‘machine learning’, ‘UAV’, ‘unmanned aerial vehicle’, and ‘sugarcane’ within the searches undertaken in Google Scholar, Scopus, and Web of Science. Logical combinations of these words were also used to identify appropriate phrases within titles, abstracts, and keyword summaries. Eighteen journal articles were chosen and analysed. These were separated into seven research segments associated with sugarcane production. They included yield prediction, fertilizer monitoring, disease monitoring, irrigation monitoring, varieties and breeding, crop class, and crop quality. This article reports on the strengths, weaknesses, opportunities, and threats (SWOT) analysis used to diagnose the benefits, limitations, potential, and barriers to use of ML for sugarcane research in Thailand. This technique has an ability to determine and characterise practices and products for improving decision-making and sustainable production on-farm. However, there is a need for collaboration between government policymakers and sugar stakeholders to provide researchers with appropriate amounts of quality datasets required for calibrating and validating predicted values. ML provides opportunities but also challenges for researchers to describe spatial data effectively and appropriately from UAV for use in precision agriculture (PA) in Thailand.