Data annotation platform with control over cost, quality and data
ImageLabeling is a data annotation platform for companies that train computer vision models. This page is for the person who signs the budget: how seats and roles work, how quality is proven, how cost per labeled image is reasoned about, how your data is protected and what each plan costs. The tool itself is covered on other pages. Here we cover control.
Seats and roles that match your org chart
Every workspace has five roles. The owner controls billing and the workspace. Admins create projects, schemas and members. Reviewers approve or send back work. Labelers label. Viewers see results without editing. Access is set per project, so a contractor labeling one dataset never sees another.
- Starter 3 seats, Team 10, Business 30, Enterprise unlimited seats
- Project level access for internal teams and contractors
- Viewer role for stakeholders who only need to see progress
A QA trail you can show to anyone
Every label has a history: which labeler drew it, whether it came from an AI draft, which reviewer approved it, and what comments were exchanged when it was sent back. When a model behaves oddly on a class, you can trace the training labels for that class to the people and decisions behind them. On Team and higher, consensus on a sample of images shows where labelers disagree. On Business, a QA score per labeler turns that into a number you can manage.
- Who labeled what, who approved it, and when
- Consensus on a sample of images (Team and higher)
- QA score per labeler (Business and higher)
- Frozen dataset versions tied to the labels each model saw
How a data annotation platform lowers cost per labeled image
The cost of a labeled image has three parts: people time, review time and rework. A data annotation platform lowers each one in a different way. AI pre-labels replace drawing with checking, which is where most people time goes. A clear reviewer workflow catches errors once, before training, instead of after a failed model run. Frozen versions and version compare stop teams from relabeling data they already paid for. To estimate your own number, take one batch, measure labeler hours with AI drafts on, add reviewer hours, and divide the plan price plus those hours by the approved images. Run the same batch next month and the trend tells you whether QA rules are working.
- Plan price is flat per month, no credits and no usage meter for manual labeling
- AI quota covers images pre-labeled once, re-exports never count
- Deleting images never frees the monthly AI quota, so the number you see is the number you used
Security, SSO and audit log
Images and labels are encrypted in transit and at rest. Your images are used only to label your dataset and are never used to train third party models. You can request deletion of your data. Business adds Google and Microsoft SSO and an audit log of workspace actions. Enterprise adds SAML SSO, a data processing agreement and a 99.9 percent uptime SLA. Details are on the security page.
- Encryption in transit and at rest
- Your images never train third party models
- Audit log and Google or Microsoft SSO on Business
- SAML SSO, DPA and 99.9 percent uptime SLA on Enterprise
Billing that procurement understands
All plans are subscriptions, billed monthly or yearly. Yearly billing saves 50 percent and invoices come with yearly plans, with invoicing on Enterprise for finance teams that pay that way. You can cancel anytime and access runs to the end of the paid period. Payments are final.
Managed labeling when you lack people
On Business and Enterprise you can add vetted annotators from 899 USD per month per dedicated annotator. They work inside your workspace, under your schema, your reviewers and your QA rules, so quality data stays in one place. See data annotation services for how annotators are matched.
Plans at a glance
The full comparison is on the pricing page. Feature differences per plan are laid out on the image annotation software page, and the day to day steps on the image annotation workflow page.
| Plan | Monthly | Yearly per month | Seats | AI pre-labeled images per month | Stored images |
|---|---|---|---|---|---|
| Starter | 49 USD | 24 USD | 3 | 1,000 | 10,000 |
| Team | 149 USD | 74 USD | 10 | 3,500 | 100,000 |
| Business | 499 USD | 249 USD | 30 | 12,000 | 1,000,000 |
| Enterprise | 1,499 USD | 749 USD | unlimited | 35,000 | 5,000,000 |
FAQ
Who owns the labels and images?
You do. Export them anytime in COCO, YOLO, Pascal VOC or CSV. Your images are used only to label your dataset.
Do we need a sales call to buy Enterprise?
No. Every plan, Enterprise included, has its price on the page and self serve checkout.
Can we start small and move up later?
Yes. Start on Starter or Team and change plans as seats or volume grow. Moving to a higher plan is charged right away for the difference.
Is there an audit log on every plan?
The audit log comes with Business and Enterprise, together with SSO.
Put your labeling budget under control
Create a workspace, invite your team and label your first batch with AI drafts and review on.