Age Estimation API reference

The Rigr AI facial age estimation API accepts images and returns an estimated age for every detected face, with a bounding box, a detection confidence score and a calibrated uncertainty value. The current model (v3) achieves a mean absolute error of approximately 1.6 years on our benchmark dataset.

It runs as a hosted cloud service or as a self-hosted Docker container for on-premise deployment, which can be air-gapped. You can get a free API key to evaluate it, read the capability overview for what the model does and where it is used, or download this reference as a PDF.

Rigr AI took part in the Australian Age Assurance Technology Trial (2025), an independent evaluation commissioned by the Australian Government and conducted by the Age Check Certification Scheme against a dataset the trial built. Its assessment found Rigr strong on technical readiness, privacy alignment and usability, with low false-positive rates in real-world scenarios.

Quick start

The hosted cloud API is available at https://api.age.rigr.ai.

# Health check (no auth required)
curl https://api.age.rigr.ai/healthz

# Verify your API key and check server info
curl -H "X-API-KEY: YOUR_KEY" https://api.age.rigr.ai/api/info

# Estimate age from an image
IMG_B64=$(base64 -w0 photo.jpg)
curl -X POST https://api.age.rigr.ai/api/image \
  -H "Content-Type: application/json" \
  -H "X-API-KEY: YOUR_KEY" \
  -d "{\"images\": [\"${IMG_B64}\"]}"

The cloud API may have a cold-start delay of a few seconds on the first request if the container has scaled to zero. Subsequent requests are fast. For a self-hosted deployment, replace the host with your server address (for example http://localhost:8000).

Authentication

Every endpoint except /healthz requires an API key in the X-API-KEY header. Keys are issued by Rigr AI — request one through the self-serve signup.

X-API-KEY: your_api_key_here

Endpoints

GET /healthz

Lightweight health check that verifies the server is reachable. Requires no authentication and does not wake the container on the cloud deployment, so it is safe for uptime monitoring. Returns HTTP 200 when the service is healthy.

POST /api/image

Submit one or more base64-encoded images for age estimation.

Each detected face is returned as a FaceResult:

Example response:

{
  "error": null,
  "results": [
    {
      "idx": 0,
      "error": null,
      "results": [
        { "idx": 0, "age": 25.3, "uncertainty": 1.2,
          "bbox": [175, 133, 364, 378], "score": 0.9998, "source": "" },
        { "idx": 1, "age": 10.5, "uncertainty": 2.1,
          "bbox": [633, 234, 725, 347], "score": 0.9993, "source": "" }
      ]
    }
  ]
}

GET /api/info

Returns the current pipeline configuration — model version, batch and image limits, and the detector and estimator models and devices in use. Requires authentication.

{
  "version": "v3",
  "max_batch_size": 20,
  "max_img_mb": 50.0,
  "min_img_px": 24,
  "max_img_px": 6000,
  "detector_model": "retinaface_resnet50",
  "detector_device": "cuda",
  "estimator_device": "cuda",
  "estimator_model": "rigr_age_v3"
}

GET /api/stats

Returns throughput statistics for the detection stage, the estimation stage and the combined pipeline — average task time, items per second, totals processed and task count. Requires authentication.

Image requirements

Error handling

Errors are structured JSON objects with type, message and detail fields, returned at two levels: a top-level error that prevented the whole request from being processed, and a per-image error affecting one image while the rest are still processed. A per-image error can appear alongside results — a truncated image is still processed, with the error field flagging the issue.

A request body that does not conform to the expected schema returns HTTP 422 with a RequestValidationError listing each offending field.

Deployment

Cloud (evaluation). The fastest way to evaluate is the hosted instance at https://api.age.rigr.ai — no infrastructure setup required. The cloud deployment runs the lite (ONNX-only, CPU) container.

Self-hosted and air-gapped. Pre-built Docker images support CPU and GPU (CUDA) inference. All images expose the same API on port 8000, so only the host changes. Contact us for registry access, image tags and deployment guidance.

Getting access

Get a free API key Talk to us about deployment

Also available: the Age Estimation capability overview, the Trust & Safety solution page for review-routing workflows, and this reference as a PDF.