# Ultralytics YOLO

> State-of-the-art open-source YOLO models for object detection, instance segmentation, image classification, pose estimation, and tracking.

Ultralytics YOLO is a cutting-edge, open-source computer vision framework built on years of foundational research in AI and deep learning. It provides a family of fast, accurate, and easy-to-use YOLO models that excel at object detection, instance segmentation, image classification, pose estimation, and oriented bounding box detection. The library is installable via pip and supports both a CLI and a Python API, making it accessible to researchers and production engineers alike. It is licensed under AGPL-3.0 for open use, with an Enterprise License available for commercial deployments.

- **Object Detection** — *Run pretrained YOLO26 models on images or video streams with a single `yolo predict` command or Python call.*
- **Instance Segmentation** — *Use YOLO26-seg variants to produce pixel-level masks for each detected object.*
- **Image Classification** — *Classify images across 1000 ImageNet classes using YOLO26-cls models pretrained for high top-1 accuracy.*
- **Pose Estimation** — *Detect human keypoints with YOLO26-pose models trained on COCO-Pose.*
- **Oriented Bounding Boxes (OBB)** — *Detect rotated objects in aerial imagery using YOLO26-obb models trained on DOTAv1.*
- **Object Tracking** — *Apply multi-object tracking across video frames; compatible with all detection, segmentation, and pose models.*
- **CLI & Python API** — *Install with `pip install ultralytics`, then use the `yolo` CLI or import `from ultralytics import YOLO` in Python.*
- **Multiple Export Formats** — *Export models to ONNX, TensorRT, CoreML, TFLite, and more for cross-platform deployment.*
- **Integrations** — *Native support for Weights & Biases, Comet ML, Roboflow, Intel OpenVINO, Neural Magic DeepSparse, and the Ultralytics Platform.*
- **Multi-language Docs** — *Full documentation available in Chinese, Korean, Japanese, Russian, German, French, Spanish, Portuguese, Turkish, Vietnamese, and Arabic.*

## Features
- Object detection
- Instance segmentation
- Image classification
- Pose estimation
- Oriented bounding box detection
- Multi-object tracking
- CLI and Python API
- Model export (ONNX, TensorRT, CoreML, TFLite)
- Pretrained YOLO26 models
- Custom model training
- Conda and Docker support

## Integrations
Weights & Biases, Comet ML, Roboflow, Intel OpenVINO, Neural Magic DeepSparse, Ultralytics Platform, PyTorch, ONNX, TensorRT, Google Colab, Kaggle, Paperspace Gradient, Amazon EC2

## Platforms
LINUX, WEB, API, DEVELOPER_SDK, CLI

## Pricing
Open Source, Free tier available

## Version
v8.4.47

## Links
- Website: https://docs.ultralytics.com
- Documentation: https://docs.ultralytics.com
- Repository: https://github.com/ultralytics/ultralytics
- EveryDev.ai: https://www.everydev.ai/tools/ultralytics-yolo