YOLO

YOLO (You Only Look Once) is a state-of-the-art, single-stage object detection algorithm: it processes an entire image in one pass for real-time performance.

YOLO, originally developed by Joseph Redmon et al. in 2015, revolutionized computer vision by reframing object detection as a single regression problem. This architecture uses one Convolutional Neural Network (CNN) to simultaneously predict bounding box coordinates and class probabilities, eliminating the multi-step pipelines of prior models like R-CNN. This unified approach delivers exceptional speed: the base YOLOv1 model achieved 45 Frames Per Second (FPS). Modern iterations, such as Ultralytics' YOLOv8 and YOLOv11, continue this legacy, offering industry-leading accuracy (mAP) and efficiency for applications like autonomous driving and video surveillance.

https://docs.ultralytics.com/