Documentation — CoreLabel

Platform overview

CoreLabel Datum is an async data annotation pipeline. You upload a dataset, the platform pre-labels it with AI, routes low-confidence items to human reviewers, runs quality checks, and delivers a structured export.

The pipeline runs entirely in the background — submit a job and come back when it's done. There's no manual labeling UI; CoreLabel is the orchestration layer between your raw data and your annotation output.

Quickstart

1

Create a project — go to Projects → New project. Pick a data type (image, video, text, audio, or structured) and give it a name.

2

Upload your data — go to Upload. Drag your files into the drop zone, or connect an existing S3/GCS bucket.

3

Submit a job — after upload completes you'll land on New Job with your data pre-filled. Select your project and hit Submit.

4

Monitor progress — watch the job move through pipeline stages on the Jobs page. The progress bar and status update in real time.

5

Download results — when status shows Completed, click ↓ Data to download your annotated dataset.

💡 First time? Start with 10–50 files to validate the output format before running your full dataset.

Projects

A Project groups jobs that share the same data type and label schema. Create one project per annotation task — for example, one for "product images" and a separate one for "customer reviews".

Data types

Type	Supported files	AI models used
Image	jpg, jpeg, png, webp, tiff, bmp	SAM2 + Grounding DINO
Video	mp4, mov, avi, mkv, webm	SAM2 (frame-by-frame)
Text	txt, csv, jsonl, json, pdf, docx	spaCy + BART + DeBERTa
Audio	wav, mp3, flac, ogg, m4a	Whisper + classifier
Structured	csv, json, jsonl, parquet, arrow	Custom model

You cannot change the data type after a project is created. If you need a different type, create a new project.

Uploading data

Go to Upload and drag files into the drop zone, or click to browse. Files upload directly to secure object storage — the backend never proxies your bytes, so large files are fast.

File-type validation

The allowed extensions depend on the data type you select. Rejected files show a ✗ icon and are skipped.

Progress tracking

Each file shows its own status icon: ⏳ uploading → ✓ done or ✗ failed. The overall progress bar updates as files complete. When all files finish, you're automatically redirected to New Job with the storage prefix pre-filled.

⚠ Uploaded files that are never turned into a job are automatically deleted after 7 days.

Cloud connector

Instead of uploading files, you can link an existing cloud bucket. CoreLabel reads data in-place — your files never leave your storage. Only annotation outputs are written back to CoreLabel's storage.

AWS S3

Switch to the Cloud Connector tab on the Upload page.
Select AWS S3 and enter your bucket name.
Enter an Access Key ID and Secret Access Key for an IAM user with s3:ListBucket and s3:GetObject permissions on that bucket.
Click Verify connection and wait for the status to turn green.

Google Cloud Storage

Select Google Cloud Storage and enter your bucket name.
Paste the full JSON contents of a service account key file. The service account needs the Storage Object Viewer IAM role.
Click Verify connection.

💡 Credentials are encrypted with AES-256 (Fernet) before being stored. They are only decrypted when a Celery worker needs to access your bucket.

Verification

Verification lists your bucket to confirm access and counts the files. This typically takes 5–30 seconds. The connector must be in verified status before you can create a job with it.

Jobs & pipeline

A Job is one annotation run. It takes a dataset and a project, then moves through the following pipeline stages automatically.

Pipeline stages

Stage	What happens
Ingesting	Files are indexed; item count confirmed
Queued	Waiting for an AI worker slot
Pre-labeling	AI model generates candidate labels for each item
Annotating	Low-confidence items routed to human reviewers Growth+
QA review	Quality checks run; IAA score calculated
Assembling	Export file and quality report built
Completed	Output ready to download
Failed	Pipeline error — check the audit log

Human review routing

Items where the AI confidence score falls below 0.75 are automatically escalated to human annotators, regardless of your tier. On Growth and above, human review is fully enabled for all flagged items. On Starter, low-confidence items are re-run with a fallback model instead.

IAA score

Inter-Annotator Agreement (IAA) measures how consistently different annotators labeled the same items. It ranges from 0.0 (no agreement) to 1.0 (perfect agreement).

Score	Interpretation
≥ 0.80	Excellent — label guidelines are clear
0.60 – 0.79	Acceptable — review ambiguous cases
< 0.60	Low — revisit your label definitions before proceeding

If IAA is consistently low, consider adding more examples to your label guidelines or reducing the number of label classes.

Quota

Each data item in a job consumes one quota point. Quota is tracked per calendar month (UTC) and resets on the 1st. Unused quota does not roll over.

Check your current usage on the Quota page at any time. When you submit a job, quota is reserved upfront. If the job fails, quota is not automatically refunded.

⚠ If you hit your limit mid-month you'll see a quota-exceeded error (HTTP 429) on job submission. Upgrade your plan or wait for the monthly reset.

Plans & tiers

Feature	Starter	Growth	Scale	Enterprise
Monthly quota	500	5,000	50,000	Unlimited
Human review	—	✓	✓	✓
Priority queue	—	—	✓	✓
Webhooks	—	—	✓	✓
Export formats	JSON, CSV	JSON, CSV	+ COCO, YOLO	+ Pascal VOC, custom
Seats	1	2	10	Unlimited
SLA	—	72 h	24 h	8 h
VPC / on-prem	—	—	—	✓

Billing

Go to Billing to upgrade your plan. Payments are processed securely through Stripe — CoreLabel never stores your card details.

After a successful payment your tier is updated immediately; you don't need to log out or refresh. To downgrade, contact support — downgrades are handled manually to prevent accidental data loss.

Export formats

When a job completes, click ↓ Data on the Jobs page to download the annotation export in your project's configured format — JSON, CSV, COCO, YOLO, or Pascal VOC. The export contains every item with its labels, confidence scores, annotator metadata, and audit trail.

JSON structure

{
  "job_id": "abc123",
  "project": "Product images",
  "completed_at": "2026-06-28T12:00:00Z",
  "iaa_score": 0.87,
  "items": [
    {
      "id": "item_001",
      "source_key": "uploads/abc123/cat.jpg",
      "labels": [
        { "class": "cat", "confidence": 0.94, "bbox": [10, 20, 200, 180] }
      ],
      "reviewer": "human",
      "annotated_at": "2026-06-28T11:55:00Z"
    }
  ]
}

Other formats

COCO, YOLO, and Pascal VOC export formats are available on Scale and Enterprise plans. Custom schema export is available on Enterprise. Select the format when creating the project.