Helmets Labeling Crops

University of Maryland/ NASA Harvest Africa Program led a diverse partnership to create an unprecedented, labeled datasets covering the main food production regions in 5 African Focus Countries of Kenya, Mali, Rwanda, Tanzania, and Uganda. This team of experts leveraged expertise, experience, and networks of universities across the continent to create opportunities for training and exposing future African researchers to concepts of remote sensing and machine learning, and to scale this approach to new partnerships in other Focus Countries. This project implemented a novel and innovative approach that utilized rapid point data collection with the “Helmets,” combined with crowdsourcing (the student network), to generate point and polygon labels using tested software tools to create an ML ready label dataset hosted on Radiant Earth ML Hub. The dataset is now being tested with on-going projects led by or coordinated by project partners.  The data includes crop type labels (points and polygons), crop pest and diseases, and market prices. Drawing on years of experience our approach prioritizes data quality, efficiency, minimizes cost per label and ensures the safety of our field teams.