AnnotateIt AI
Offline image annotation, without the cloud
Local-first computer vision annotation for ML engineers, researchers and small CV teams.
Your images never leave your device.
AnnotateIt AI is a private, local-first workstation for labeling computer vision datasets. You annotate images and video frames on your own machine — with AI assistance — and export straight to the formats your training pipeline already speaks. No account, no upload, no subscription.
Why annotate offline
- Private by design. Datasets, media and annotations stay on your device — nothing is uploaded to a third-party annotation service.
- Works offline. Keep labeling without a network connection; desktop and mobile builds bundle their AI models.
- No account required. Open the app and start a project — there is no sign-up and no cloud project backend.
- No subscription. The core workflow is free; Pro is a one-time purchase, not a recurring bill.
What you can annotate
- Object detection with bounding boxes
- Instance segmentation with polygons and masks
- Single-label, multi-label and hierarchical classification
- Keypoint detection
- Video: extract frames and annotate them like images
AI-assisted, on your device
Smart tools speed up labeling without sending your pixels anywhere. Segment objects with Segment Anything (SAM), refine with interactive tools, and review AI pre-annotations before accepting them. The models run locally, so assistance works the same whether or not you are online.
Export to standard formats
Import to fix or extend labels, export to train. Datumaro preserves the most on a round trip — in COCO, YOLO and Pascal VOC anything that is not a polygon is written as its bounding box.
| Format | Typical use |
|---|---|
| YOLO | Detection and segmentation training with the YOLO family |
| COCO | Detection, segmentation and keypoints across common frameworks |
| Pascal VOC | Classic per-image XML detection annotations |
| Datumaro | Universal JSON format for dataset conversion and management |
How it works
- Create a project — choose detection, segmentation, classification or keypoints.
- Import media — add images, or extract video frames into the dataset as images.
- Annotate faster — use manual tools, Segment Anything (SAM) and AI pre-labeling.
- Export your dataset — YOLO, COCO, Pascal VOC or Datumaro.
Built for sensitive and proprietary data
Local-first annotation fits workflows where images cannot — or should not — be uploaded to a shared cloud:
- Medical and clinical research imagery
- Manufacturing and industrial inspection
- Security and surveillance footage
- Robotics, agriculture and edge-AI datasets
- Confidential client data handled by independent ML engineers and consultants
AnnotateIt vs. cloud annotation tools
| AnnotateIt AI | Typical cloud tool | |
|---|---|---|
| Where your images live | On your device | Uploaded to a vendor cloud |
| Account required | No | Usually yes |
| Works offline | Yes | No |
| AI assistance | On-device Segment Anything (SAM) | Server-side |
| Pricing | Free, then one-time Pro | Recurring subscription / per-seat |
Pricing
Free for up to 5 active projects, with the full annotation workflow, on-device AI and import/export included. AnnotateIt Pro is a one-time $49 lifetime upgrade that removes the project limit — no subscription. Your projects always remain exportable and deletable, whichever tier you are on.
Frequently asked questions
Does AnnotateIt upload my images to the cloud?
No. Your projects, images, video frames, labels and annotations are stored and processed on your own device. The web build downloads the app and AI model files over the network, but it does not upload your project content.
Can I annotate without an internet connection?
Yes. The desktop and mobile apps bundle their AI models and run fully offline. Once the web build has loaded, annotation and on-device AI keep working without a connection.
Which platforms are supported?
The browser app, plus native builds for Windows 10/11, iPhone and iPad, and Apple silicon Macs.
Start annotating locally.