CVAT

Leading open-source annotation platform for image, video, and 3D vision

Editorially reviewedChecked Sep 2026How we review

mlops#annotation#computer-vision#image-labeling#open-source

Free plan Free trial API Self-hosted Teams

Toolglade’s take

Best as a mature open-source annotation tool for images and video, with strong support for bounding boxes, polygons, and segmentation you can self-host for free. The caveat is setup and scaling take real effort, and the UI feels utilitarian next to polished commercial rivals.

Should you use CVAT?

Best for: Computer-vision teams · ML data ops · Researchers
Not ideal for: Text-only NLP labeling · Non-technical solo users wanting turnkey SaaS · Audio annotation needs
Biggest strength: Free and open source (MIT) self-hosted
Biggest weakness: Self-hosting requires setup and maintenance
Pricing: Free plan, paid from $0
Top alternative: Labelbox · see all

Facts & pricing last checked September 2026. How we review

About CVAT

CVAT is a leading open-source annotation platform for image, video, and 3D data, with AI-assisted labeling, QA, and analytics, available free self-hosted or via paid cloud tiers.

CVAT is the Computer Vision Annotation Tool, a widely adopted open-source platform for annotating images, video, and 3D data for computer-vision projects that require high precision and scale. Originally developed by Intel and now maintained by CVAT.ai, it has grown into one of the most widely used annotation platforms, with a large global developer community building on it. CVAT handles nearly every computer-vision annotation task: bounding boxes, polygons, polylines, keypoints, semantic segmentation, video tracking, and 3D cuboids for LiDAR and point-cloud data. It adds AI-assisted labeling, quality assurance, team collaboration, analytics, and developer APIs, making it suitable for both individual researchers and larger data-operations teams. The self-hosted version is fully free and open source under the MIT license, so teams with their own infrastructure can run it without cost, while paid cloud and enterprise tiers offer managed hosting and additional capabilities. Active development continues, with regular releases keeping the tool current for modern vision AI workflows.

Weighing your options?See how CVAT compares to the alternatives.

TL;DR

CVAT is a leading open-source annotation platform for image, video, and 3D vision data, free to self-host with AI-assisted labeling, QA, and paid cloud tiers.

Company overview

CVAT began as an Intel project and is now maintained by CVAT.ai. It has become one of the most widely adopted open-source annotation platforms, used by a large global community to build visual datasets for vision AI.

The company offers open-source, cloud, and enterprise products plus labeling services, spanning individual researchers to large data-operations teams. Development is active, with regular releases keeping it current.

Product features

CVAT supports the full range of computer-vision annotation tasks, from bounding boxes and polygons to keypoints, segmentation, video tracking, and 3D cuboids for LiDAR. It adds AI-assisted labeling, QA, team collaboration, analytics, and developer APIs.

The self-hosted Community edition is free under MIT, while cloud and enterprise tiers provide managed hosting and additional capabilities. Integrations and SDKs let teams fit CVAT into automated data pipelines.

Target market

CVAT targets computer-vision teams, ML data-operations groups, researchers, and organizations working on autonomous and robotics data. It is less suited to text-only NLP or audio annotation needs.