Artificial intelligence-assisted drone approach for accurate stand count in Moroccan sugar beet fields

Recent advances in artificial intelligence (AI) offer a promising alternative to conventional drone-based remote sensing.In this study, we propose a sustainable stand-enumeration methodology for Moroccan sugar beet fields by integrating high-resolution imagery acquired with unmanned aerial vehicles (UAVs) and a novel deep-learning pipeline.Four convolutional-and transformer-based architectures YOLOv5, Fast R-CNN, YOLOR (YOLOR) and YOLOv7 were trained and evaluated on imagery from the Beni Mellal region using precision, recall, mean average precision (mAP) and plant survival-rate metrics.YOLOv5 achieved the best performance, with 97% precision, 92% recall and 96% mAP, significantly outperforming the other models.These results demonstrate that AI-assisted UAV imagery can deliver highly accurate crop-stand counts, thereby supporting data-driven decision-making and enhancing sustainability in agricultural monitoring.