Images
- Detect objects with bounding boxes
- Segment scenes with pixel-level masks
- Segment instances with polygons
- Estimate poses with keypoints & skeletons
Cloud Data Annotation Platform for Vision & Audio AI
Turn raw images, videos, 3D point clouds and audio into model-ready datasets. Run the full workflow from dataset distribution to AI-assisted labeling to quality review, all in one managed workspace.
Trusted by ML teams worldwide
Read customer stories- FERNRIDE
- TB Maestro
- MiR
- EPAM
- Gendo
- Purdue University
- Vivint
- Arbo Air
- Kombai
- Diligent
- UIC College of Medicine
- Tecnalia
- Vimaan
- Dragonfly AI
All-in-one data annotation platform for AI teams
Videos
- Track any object across every frame
- Track skeleton graphs through movement
3D point clouds
- Detect and classify objects in 3D spaces
- Track objects through point cloud sequences
Audio
- Transcribe multilingual speech into time-aligned text
- Detect and segment speech and sound events
Full-featured annotation editor for simple and complex vision tasks
Create accurate labels with dedicated shapes, modes, and editing controls for classification, segmentation, and tracking.
Multimodal data support
Annotate images, videos, 3D LiDAR point clouds, and audio directly in your browser.
Diverse labeling shapes & tools
Use the right annotation type for the task, from bounding boxes and polygons to pixel masks, skeletons, and 3D cuboids.
Multiple annotation modes
Switch between dedicated modes for standalone objects, sequences, frame-level classes, and object attributes.
Advanced shape and track editing
Refine and restructure annotations without redrawing them using snap, merge, group, split, join, and slice tools.
Video tracking and interpolation
Track objects across frames with automatic shape interpolation between keyframes.
Purpose-built tools for 3D data
Place, resize, rotate, and track cuboids in point clouds with projection views for precise positioning and orientation.
Bulk shape conversion
Convert masks, polygons, and rectangles in bulk without recreating existing annotations.
Layers, filters, and canvas controls
Organize the canvas to match your workflow with layers, object filters, visibility controls, image adjustments, and customizable shortcuts.
AI-assisted labeling with built-in and custom models
Speed up annotation by pre-labeling data with native models, models from external hubs, or your own models.
Segment matching objects with SAM 3
Quickly segment all matching objects in a frame with Segment Anything 3 using a text or visual prompt.
Track objects across frames with SAM 2
Track objects across frames automatically with Segment Anything 2 instead of recreating them frame by frame.
Bring the right pretrained model for the task
Connect any compatible model from supported hubs to pre-label data for your specific use case.
Connect your own models through AI Agents
Generate first-pass annotations with models trained on your own data.
Organized annotation workflows and team collaboration
Easily organize, manage, and review annotation projects across parallel workstreams for in-house or outsourced annotators.
Structured workflows for annotation projects
Manage the full workflow of setting up projects, splitting datasets, assigning jobs, reviewing, and approving annotations.
- Define labels and attributes once at the project level and apply them across every task and job.
- Automatically split large datasets into manageable jobs and distribute them for parallel annotation.
- Track assignees, workflow stages, status, and completion for every project, task and job.
Team collaboration
Highlight and discuss annotation issues, share project instructions, and align on labeling decisions.
Secure role-based controls
Keep your data safe through organization and project-level roles.
Team activity audit trail
Monitor user activity and maintain a traceable record across projects, tasks, and jobs.
Advanced data annotation QA
Catch labeling errors before they reach your training set with a comprehensive set of manual and automated validation tools.
Benchmark jobs against Ground Truth.
Automatically spot missing or extra annotations and label mismatches by comparing regular jobs against validated Ground Truth jobs.
Resolve ambiguity through consensus.
Reduce individual bias in ambiguous cases by automatically comparing independent annotations and flagging disagreements for review.
Review and resolve annotation issues.
Use Review mode to open issues on specific objects or frame regions, discuss corrections, and return jobs for rework.
Immediate Job Feedback.
Give job assignees an instant quality score upon completion, so below-threshold work can be corrected immediately.
Annotation analytics and reporting
Get insights into your labeling team's performance. Spot bottlenecks and balance workloads to keep dataset delivery on track.
Measure time and annotation speed
Track total working time and average annotation speed to estimate effort and understand labeling throughput.
Find workflow bottlenecks
Break down activity by assignee, workflow stage, job status, and date to see where work slows down or waits for the next handoff.
Analyze annotation output
Review object counts by label and annotation type, along with tracks, keyframes, and interpolation rates, to understand dataset composition and labeling progress.
Export data for deeper analysis
Export annotation statistics and event data for custom reporting and analysis.
USA
“CVAT is central to our annotation pipeline for two reasons: SAM-powered labeling that saves critical time for busy surgeons, and multi-user access that enables seamless collaboration for researchers across continents.”
Read case study
Luca Morgantini
MD Resident, Department of Urology
CANADA
“The reason I chose CVAT is because whenever I had a question or an issue, I got a quick response from the team. Unlike other tools, the team behind CVAT was always listening and willing to work with me. The labeling platform is great and user-friendly, but it's the team behind it that makes the difference.”
Read case study
Véronica Romero-Rosales
Co-founder, Robonotic
GERMANY
“I love CVAT because it's so hands-on, simple and straightforward. We tried other well-known tools before, but there was a lot of friction when we tried to get them up and running, and the UI was harder to work with.”
Read case study
Andre Kempe
CEO & Product Owner, ProMetronics
NETHERLANDS
“The biggest win for us is the model-assisted annotation loop. We run YOLO and SAM2 pre-annotation directly through CVAT Online, so the work is mostly verifying and correcting rather than labeling from scratch. For a small operation producing the computer-vision training data behind our product, that's the difference between a dataset taking weeks versus days.”
Read case study
Fabian van Schevikhoven
Co-founder, Younalize
COMMUNITY VOICE
“We have a dedicated annotation team within our company, comprising over 50 annotators. For the past four years, we have been using self-hosted CVAT, which has been functioning exceptionally well. Recently, we acquired a project that requires annotating approximately 1 million images and videos monthly. We tried various tools, such as Supervisely, Label Studio etc, especially for video annotation, but CVAT remains the best option.”
Built to fit into your ML pipeline, not the other way around
Connect CVAT to your existing stack and automate annotation workflows without replacing your current tools or infrastructure.
Automate workflows with the API and SDK
Create projects and tasks, upload data, monitor jobs, and retrieve annotations programmatically.
Script repeatable workflows with the CLI
Script project and task creation, dataset import and export, backups, and auto-annotation from a terminal or CI pipeline.
Trigger downstream steps with webhooks
Send project, task, and job events to external systems for notifications, handoffs, and downstream automation.
Connect your cloud storage
Import and export datasets through Amazon S3, Azure Blob Storage, Google Cloud Storage, and S3-compatible services.
Export in 20+ computer vision formats
Deliver finished annotations in COCO, YOLO, KITTI, Cityscapes, Pascal VOC, and other formats used by downstream ML tools.
Use prebuilt integrations
Connect CVAT to supported model hubs, dataset tools, and crowdsourcing workflows without building every connection from scratch.
Plans for individuals and teams
Start free and upgrade as your annotation workload or team grows.
Free
$0
For exploring CVAT and small projects
What`s included:
- Core annotation editor
- Manual review
- Ground Truth and Honeypot QA
- API access
- Annotation export
Solo
From$23/month
For individual professionals with ongoing labeling work
Everything in Free, plus:
- Full dataset export
- Analytics and reporting
- Batch auto-annotation
- SAM 2 and SAM 3 image segmentation
Team
From$23/user/month
For in-house teams and outsourced labelers
Everything in Solo, plus:
- Shared organization workspace
- Team assignments and reviews
- Role-based access controls
- SSO and audit logs
Why AI teams choose CVAT for labeling their data
4.5/5“CVAT helped me build an AI-powered inventory counting solution for warehouse operations. The annotation workflow was efficient, and it made it much easier to create high-quality training data for accurate object detection and counting.”
Sukanya Suman Kalyan
Azure Data Engineer, Deloitte
“We selected CVAT after evaluating several options and determined it was the best fit for our instance segmentation workflows. Features such as automatic bordering and bitmap visualization help us improve annotation quality while maintaining high productivity.”
Jorge Rojas
Ph.D. Candidate & Research Assistant, Purdue University
“SIYTE is an AI-powered intelligence platform connecting cameras, sensors and systems with CVAT playing an instrumental role in supporting our data annotation and model training function. CVAT is a reliable and easy to use tool that genuinely supports how our teams build and refine computer vision models.”
Jamie Gale
Chief Technology Officer, SIYTE
“At Dragonfly AI we help brands measure creative effectiveness of marketing materials e.g. packaging, advertisements etc. using patented AI. CVAT helps us efficiently track attention-critical areas in video ads and feed structured annotation data into our proprietary attention measurement algorithm. The platform is straightforward, reliable, and fits naturally into our workflow.”
Will Champion
Customer Success Executive, Dragonfly AI
“We began our AI journey with CVAT because, at the time, it was the only solution truly bridging the gap between raw data capture and practical model training workflows. Years later, it continues to be a foundational part of our pipeline thanks to its robustness, flexibility, and deep alignment with real-world computer vision development needs.”
Maikel Boerebach
Business Development Manager, AKL
“We chose CVAT for its strong balance of capability and cost control. It allows us to build and manage our annotation workflows without unnecessary overhead.”
Craig Metcalfe
COO/CPO, Illuminex AI
Explore more from CVAT
CVAT Online vs. alternatives
Learn and stay up to date
Explore other CVAT solutions
Frequently Asked Questions
How much does CVAT Online cost?
CVAT Online offers a Free, Solo, and Team plans.
- The Solo plan is $33/month or $23/month billed annually (30% off).
- The Team plan starts at $66/month or $46/month if billed annually (minimum 2 users). Additional seats are $33 per user/month or $23 per user/month billed annually (30% off).
See the full pricing and feature limits (projects, tasks, storage, API/export, security options like SSO/RBAC) on the pricing page.
For larger organizations that need private hosting and advanced controls, CVAT Enterprise is available and typically starts at $10,000/year.
Is it possible to deploy CVAT on-premises?
Yes.
CVAT offers on-premises deployment via two products:
- CVAT Enterprise, a commercial, supported edition with SSO/SAML, role-based access control, workspace/org management, and enterprise support (deployable in your data center or private cloud). Learn more.
- CVAT Community, a free, open-source edition you can self-host with your preferred stack. Learn more.
If you have a free plan, does it mean you’re selling my data somewhere else?
No.
We don’t sell your data. The free plan is only supported by feature and usage limits, e.g., AI calls, compute/bandwidth (see our Fair Usage Policy), not by data monetization.
Who can access my data when I use CVAT products or services?
CVAT Online / Enterprise: You own your content and control who you invite. We only process your data to operate and support the service. Limited, audited access may occur for support or compliance needs and only under confidentiality.
Labeling Services: Only the assigned, NDA-bound project team has access, using least-privilege principles and project-level isolation.
Do you offer professional data labeling services?
Yes.
CVAT provides professional annotation services run by our in-house team (300+ annotators across time zones) with NDAs, layered QA (manual + automated), and flexible pricing (project-based or subscription). Minimum project budgets typically start at $5,000.
Learn more.
How secure are CVAT’s products and services?
We apply technical and organizational safeguards (e.g., encryption in transit, role-based access, SSO options, access logging). Enterprise deployments add controls aligned to your environment and policies.We respect your privacy, do not sell personal information, and provide controls for access, export, and deletion as outlined in our policies:
Get Started Today
Build, scale, and deliver high-quality training data for your AI models with CVAT.
Free plan available • No credit card required • GDPR & CCPA compliant