YOLO
يولو (YOLO)
Also known as: You Only Look Once, YOLOv8, YOLOv9, YOLOv10, YOLO11, YOLO26
YOLO (You Only Look Once) is the dominant single-stage object detector family. The original YOLOv1 paper (Redmon, 2016) reframed detection as a regression problem; modern variants from Ultralytics (YOLO11, YOLO26) and Meituan (YOLOv6) extend the recipe with anchor-free heads, cross-stage feature pyramids, and unified detection-segmentation-pose-classification heads. YOLO trades a couple…
Definition
YOLO (You Only Look Once) is the dominant single-stage object detector family. The original YOLOv1 paper (Redmon, 2016) reframed detection as a regression problem; modern variants from Ultralytics (YOLO11, YOLO26) and Meituan (YOLOv6) extend the recipe with anchor-free heads, cross-stage feature pyramids, and unified detection-segmentation-pose-classification heads. YOLO trades a couple of mAP points for raw throughput — ~140 fps at 640 px on a Jetson Orin AGX. FI Tech uses YOLO11-s as the default PPE backbone for its sweet spot of accuracy and edge-friendly inference, fine-tuned on Saudi-construction footage.
