YOLO, short for You Only Look Once, is a family of object detection models that find and label objects in an image in a single fast pass, fast enough for live video.
In plain words
Older detectors reused image classifiers and checked many regions one by one, which was slow. The original 2015 YOLO paper treated detection as one prediction: a single network looks at the whole image once and outputs boxes and labels together. That let it run at 45 frames per second. Many later versions followed; the Ultralytics YOLO series now covers detection, segmentation, pose estimation and more, with YOLO26 released in January 2026.
A quick example
Point a camera at a street. A YOLO model reads each video frame and draws boxes around every car, bike and person, with labels, many times per second.
Where you'll meet it
- Real-time detection in live camera video.
- Detection, segmentation and pose estimation in photos and video.
Related
Sources
- PaperYou Only Look Once: Unified, Real-Time Object Detection, Redmon et al. (arXiv) · read 29 Sept 2026
- DocsModels supported by Ultralytics, Ultralytics Docs · read 29 Sept 2026
- DocsUltralytics YOLO Docs, Ultralytics Docs · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 31 Oct 2026. Spotted an error? Report it.