| Summary: | Attributes such as biomass and nutritional quality are characteristics that allow for the selection of better pastures and, therefore, for obtaining greater quantities of meat and milk from livestock feed. However, the laboratory analytical method used to calculate these attributes is inefficient for perennial species when there are more than hundreds of plants to evaluate. Therefore, it is necessary to develop methods that allow for the characterization and selection of genotypes using non-destructive, high-throughput techniques. For this purpose, active canopy sensors such as the Crop Circle Phenom have been used, capable of measuring reflectance, structural attributes, and climatic variables. The objective of this study was to build machine learning models to predict traits such as fresh biomass (FB), dry biomass (DBM) and nutritional traits such as crude protein (CP), in Urochloa humidicola genotypes in 3 locations (Palmira, Quilichao and Villavicencio) from harvests and analytical data (direct estimation method) and measurements acquired with portable multispectral sensors.
|
|---|