Skip to main content
CVAT (Computer Vision Annotation Tool) is an interactive video and image annotation tool designed for computer vision applications. Used by tens of thousands of users and companies worldwide, CVAT helps teams solve real problems using a data-centric AI approach.

​
What is CVAT?

CVAT is a free, open-source annotation platform that enables you to label images, videos, and 3D point clouds for training computer vision models. Whether you’re building object detection systems, semantic segmentation models, or pose estimation algorithms, CVAT provides the tools and workflows to create high-quality training datasets efficiently.

Interactive annotation

Label images and videos with an intuitive interface that supports multiple annotation types including bounding boxes, polygons, polylines, points, cuboids, skeletons, and more.

Team collaboration

Work together with your team using built-in user roles, assignment workflows, and quality control features to ensure consistent, accurate annotations.

Automatic labeling

Speed up annotation up to 10x with serverless AI functions for automatic labeling using pre-trained models or your custom models.

Flexible deployment

Choose between CVAT Cloud for instant access or self-hosted deployment for complete control over your data and infrastructure.

​
Why teams choose CVAT

​
Comprehensive annotation support

CVAT supports all major annotation formats and types:
  • Bounding boxes (rectangle, rotated rectangle)
  • Polygons and polylines for precise object boundaries
  • Points and keypoints for landmark detection
  • Cuboids for 3D object annotation
  • Skeleton annotation for pose estimation
  • Segmentation masks for pixel-level labeling
  • Tags for image-level classification

​
Enterprise-grade features

CVAT includes advanced features like quality control, analytics, webhooks, access tokens, and integration with popular ML frameworks and datasets.
  • Quality control: Built-in tools for validation, ground truth management, and quality reports
  • Organizations: Manage multiple teams and projects with role-based access control
  • Cloud storage: Connect to AWS S3, Azure Blob Storage, or Google Cloud Storage
  • Version control: Track changes and maintain annotation history
  • API and SDKs: Automate workflows using REST API, Python SDK, or CLI

​
Rich format ecosystem

Import and export annotations in 40+ formats including:
  • COCO (MS COCO Object Detection, Keypoints, Stuff)
  • YOLO (v5, v7, v8 - Detection, Segmentation, Pose, OBB)
  • Pascal VOC
  • KITTI
  • Cityscapes
  • TensorFlow Object Detection API
  • Datumaro
  • And many more…

​
Deployment options

CVAT offers flexible deployment options to match your needs:

​
CVAT Cloud

The fastest way to get started with CVAT:

Start annotating now

Sign up for free at app.cvat.ai - no installation required
Features:
  • Free tier available
  • Instant access with no setup
  • Automatic updates
  • Scale as you grow
  • Premium features with paid plans
  • Integration with Roboflow and HuggingFace models

​
Self-hosted deployment

For complete control over your data and infrastructure:
Perfect for development, small teams, or production deployments on a single server.
git clone https://github.com/cvat-ai/cvat
cd cvat
docker-compose up -d
Access CVAT at http://localhost:8080
Self-hosted deployments support SSO, LDAP, advanced analytics, and premium integrations with Enterprise support plans.

​
Use cases

CVAT powers annotation workflows across industries:
Annotate road scenes with bounding boxes, semantic segmentation, and 3D cuboids for training perception systems. Support for KITTI, Waymo, and other autonomous driving formats.
Label objects, paths, and zones in robot workspace images and videos. Create datasets for object detection, pose estimation, and scene understanding.
Annotate medical images with precise polygon and mask annotations for diagnostic AI systems. Supports DICOM and standard image formats.
Create product detection and classification datasets. Annotate fashion items, product categories, and visual attributes.
Label people, vehicles, and events in security footage. Track objects across video frames for activity recognition.
Annotate crops, pests, and field conditions for precision agriculture AI. Support for aerial imagery and drone footage.

​
Getting started

Ready to start annotating? Here are your next steps:

Quickstart

Get started in minutes with our step-by-step guide

Key concepts

Understand projects, tasks, jobs, and labels

Installation guide

Set up your own CVAT instance

​
Community and support

Join thousands of CVAT users:
  • GitHub: Report issues, request features, and contribute at github.com/cvat-ai/cvat
  • Discord: Join the community for discussions and support
  • Gitter: Ask questions and get help from the core team
  • YouTube: Watch tutorials and product demonstrations
CVAT is open-source software released under the MIT License. Contributions are welcome!

Build docs developers (and LLMs) love

Get started for freeTalk to us