Photo by Evi Radauscher on Unsplash
Wheat is one of the most important sources of human sustenance on the planet, along with rice, corn, and cassava. The ancient grain is widely studied to ensure food security and improve agricultural processes, both of which have significant implications on local and nonlocal economies. However, despite constituting one of the most cultivated cereal crops in the world, and expanding production rapidly since the 1950s, the rate of increase in wheat yields has decelerated significantly in the last couple of decades. Traditional wheat breeding and harvesting still rely heavily on manual observation and activity, which are expensive, tedious, and time-consuming tasks. To address these issues, there have been various efforts to utilize recent breakthroughs in Machine Learning and Computer Vision technologies to automate certain agricultural processes involved in Plant Phenotyping. However, one task that has still not yet been successfully automated is the calculation of wheat head density, or the counting of wheat heads. Wheat head density is a major determinant of yield potential and allows farmers to manipulate yield components in their breeding selections for successive harvests. Automation of wheat head density calculation could significantly expedite the plant phenotyping process, saving farmers time, money, and labor.