Users capture a selfie using their device camera and Yoti’s algorithm estimates their age via facial analysis. Our image capture technology captures multiple photos in a fast sequence, so if one photo is blurry or has too much glare, we still have others available to review.
An age estimate can be returned in around 1 second.
True Positive Rate (a result that correctly estimates the person as 18+) for 13 to 17 year olds correctly estimated as under 21 is 99.3%. TPR for 6 to 12-year-olds correctly estimated as under 13 is 99.5%. Regulators are most concerned with these two age ranges to ensure that under-13s and under-18s are only able to access age-appropriate goods and services.
This option should be configured with a threshold higher than your age of interest.
We are also able to show that there is no discernible bias across age, gender or skin tone for those aged between 6 and 17. Further details can be found in our Age Estimation white paper.
Images are not re-shared or re-used and are immediately, permanently deleted. Only images collected during onboarding for Yoti’s apps (in accordance with GDPR guidelines) and consented data collection exercises are used to train the algorithm.
This method is good for:
- People without ID documents
- Global coverage
- Low friction
Request body
If you wish to enable the Age estimation service, please see below:
{ "type": "OVER", "age_estimation": { "allowed": true, "threshold": 21, "level": "PASSIVE" }, "ttl": 900, "reference_id": "over_18_example", "callback": { "auto": true, "url": "https://www.yoti.com" }, "notification_url": "https://yourdomain.example/webhook", "cancel_url": "https://www.yoti.com", "synchronous_checks": true}| Parameter | Types | Description |
|---|---|---|
| allowed | true / false | Enable the verification method to be available for the user to use. |
| threshold | Integer e.g. 30 | Age threshold for under/over age limits. We recommend for this threshold to be more than the age you want to set as your barrier to entry. |
| level | NONE PASSIVE | The level of anti-spoofing for each age verification method. PASSIVE enables a passive liveness test for age estimation. |
For extra security, you can also request a liveness test (level). This is to make sure it’s a real person behind the camera, and not a 2D image, mask or bot. The technology works by processing the image(s) through a sequence of deep neural networks. Each of these examine a different element of the image to look for clues that it might not be a real person.