Biometric Age Estimate

Proof response may now include a biometric age estimate derived from the user’s facial image. This value represents a midpoint estimate of a user's age based on biometric analysis.

Location in response

The biometric age estimate is returned as part of the document data key/value list:

{
  "Key": "BiometricAgeEstimate",
  "Value": "34"
}

• Key: BiometricAgeEstimate
• Type: String (numeric value)
• Description: Estimated age of the user (midpoint), derived from biometric analysis of the submitted facial image.

Full context example

See Proof Result Data for more details.

{
  "Document": {
    "Data": [
      {
        "Key": "Page1Name",
        "Value": "United States - Georgia Driving License (2019) Real"
      },
      {
        "Key": "Page1Probability",
        "Value": "0.96807367"
      },
      {
        "Key": "BiometricAgeEstimate",
        "Value": "34"
      }
    ]
  }
}

How it works

The biometric age estimate is calculated using facial analysis models applied during the Proof process. It reflects the system’s best estimate of the user’s age at the time of capture.

• The value is a single-point estimate (not a range)
• It is not derived from document data (e.g., date of birth), but from the biometric image itself
• The estimate is independent of pass/fail signals and does not affect the overall Proof decision

Availability

The BiometricAgeEstimate attribute is:

• Returned only when a facial image is present and successfully processed

• Omitted if:

  • No selfie or facial image was provided
  • Image quality is insufficient for reliable estimation

Usage guidelines

Biometric age estimation is intended for informational and risk-assessment use cases, such as:

• Age-based routing (e.g., additional verification for edge cases)
• Fraud and anomaly detection
• Analytics and cohort segmentation

It should not be used as a sole source of truth for regulatory or compliance decisions (e.g., legal age verification).

Best practices

• Parse Document.Data into a key/value map for easier access
• Use this value in combination with document-derived data (e.g., date of birth) when available
• Apply thresholds or ranges rather than strict equality checks
• Treat the estimate as probabilistic, not definitive