 This study developed an automated system for detecting weeds in crops using UAV images. It combines digital surface models, orthomosaics, and random forest algorithms to select the most appropriate features from the images and then classify them into either weeds or crops. The accuracy of the system was tested against manually annotated data and found to be highly reliable. Additionally, the system can generate prescription maps for post-emergent weed control, which can be used to optimize crop management decisions. This article was authored by Ana Aida Castro, Jorge Torres Sanchez, Jose M. Pena, and others.