Over three agricultural years, I conducted extensive research on the principal agricultural crops in Romania, focusing specifically on the Southeast region, which represents the largest agricultural area in the country. Utilizing satellite and drone imagery throughout the agricultural seasons, I correlated classified images with ground-based data, including plant, seed, and soil samples.
I developed custom software equipped with functions to train an algorithm for automatic classification of pixels in aerial images using a large dataset of validated and calibrated examples. This effort resulted in a legend that enables automated interpretations regarding plant health, soil moisture levels, leaf density, and more. This tool is designed to assist farmers in making rapid decisions regarding irrigation, pesticide application, and other inputs.
Furthermore, I tested and utilized multiple drones equipped with various sensors available on the market, ensuring comprehensive data collection and analysis capabilities. The tool also aids in swiftly diagnosing field conditions following attacks by pests such as corn borers or damage from hailstorms. For expansive areas impractical to survey on foot, comprehensive images with automated interpretations precisely indicate the extent of damage to the land, facilitating prompt decisions on replanting in vacant zones to salvage production.
Throughout the three-year study period, I encountered instances of extreme weather causing crop damage, which were mitigated through the planting of spring crops, ultimately resulting in good yields and profitability for the land area under study.