Ultralytics

Ultralytics builds a line of fast YOLO computer vision models, plus the open-source Python package and command-line tool used to train and run them.

In plain words

YOLO models find and label things in images and video. The Ultralytics package lets you load a pretrained YOLO model, run it on images or video, train it on your own data, and export it to formats such as ONNX. Its models handle object detection, segmentation, image classification, pose estimation and tracking objects across video frames. The code is open source under AGPL-3.0, and businesses can buy a separate enterprise licence.

A quick example

The project's own example loads a small pretrained YOLO model and runs a prediction on a sample photo of a bus. From Python, the same model can be trained on a small sample dataset called COCO8 and exported to ONNX, with one call each.

Where you'll meet it

• Object detection in photos and video.
• Pose estimation and segmentation.
• Deploying vision models on edge devices.

Related

Sources

1. Docsultralytics/ultralytics, Ultralytics (GitHub) · read 29 Sept 2026
2. DocsUltralytics YOLO Docs, Ultralytics · read 29 Sept 2026

Checked by WhatIsAI against the sources above · 31 Oct 2026. Spotted an error? Report it.