CVAT
Open-source data annotation platform for images, videos, and 3D
Founded
2018
Starting Price
\u00240
Category
Developer Tools
Last updated March 19, 2026
About CVAT
CVAT (Computer Vision Annotation Tool) is an industry-leading open-source annotation platform originally developed by Intel for labeling images, videos, and 3D point clouds. Trusted by over 200,000 developers, it offers a comprehensive toolset including bounding boxes, polygons, skeletons, and cuboids with AI-assisted auto-annotation that speeds up labeling by up to 10x. Available as a self-hosted open-source edition, a managed cloud platform, or an enterprise on-premises deployment.
Pros & Cons
Pros
- Fully open source (MIT license) with no vendor lock-in — self-host for free with unlimited usage
- Industry-leading annotation toolset covering images, video, 3D point clouds, and LIDAR data
- AI-assisted auto-annotation with SAM, YOLO, and custom model support dramatically speeds up labeling
- Supports 20+ export formats (COCO, YOLO, VOC, KITTI, etc.) for seamless ML pipeline integration
- Scales from individual researchers to enterprise teams with 300+ annotators
Key Features
AI-Powered Auto Annotation
Annotate data up to 10x faster using built-in AI models like Segment Anything (SAM 2/3), YOLO, and Faster RCNN, or connect your own custom models
Comprehensive Annotation Toolset
Full suite of annotation primitives including bounding boxes, polygons, polylines, points, skeletons, cuboids, and trajectories for any computer vision task
Video & 3D Point Cloud Support
Annotate video sequences with object tracking and interpolation, plus full 3D point cloud and LIDAR annotation capabilities
20+ Export Formats
Export annotations in PASCAL VOC, COCO, YOLO, KITTI, TensorFlow Detection API, CamVid, Cityscapes, LabelMe, MOT, and many more formats
Cloud Storage Integration
Connect directly to AWS S3, Google Cloud Storage, and Azure Blob Storage for seamless data management at scale
Quality Assurance & Analytics
Built-in review workflows, ground truth jobs, honey pot validation, and performance analytics tracking annotator hours and objects per hour
Pricing
Community (Self-Hosted)
\u00240
- MIT licensed open source
- All core annotation tools
- Unlimited projects and tasks
- Docker deployment
- REST API & Python SDK
Best For
Autonomous Vehicle Development
Annotate driving footage and LIDAR point clouds with bounding boxes, polygons, and 3D cuboids for training self-driving perception models
Medical Image Analysis
Label medical imaging data (X-rays, CT scans, pathology slides) with precise segmentation masks for diagnostic AI model training
Retail & Manufacturing Quality Control
Annotate product images and video feeds to train defect detection and inventory management computer vision systems
Research & Academic Projects
Free self-hosted deployment for university labs and research teams building custom computer vision datasets with full format flexibility
