Selfie Verifications

A Selfie Verification answers the question “Is the individual present during the transaction?”. This is especially useful when combined with a Government ID Verification to associate the face of the government issued id to the individual present.

The main goals of Selfie Verifications are to ascertain the following:

Face Image Quality
1. Poses - Determine whether an individual is properly facing towards the camera
2. Quality - Determine whether the image’s resolution, lighting, and blur is sufficient for accuracy
3. Obfuscation - Determine that an individual’s portrait is not obfuscated by glasses, hair, or other items
Face Comparison
1. Determine whether the individual matches a claimed individual
Face Liveness
1. Determine whether the submission contains features that are indicative of a spoof
Age Estimation
1. Determine whether the perceived age of the individual falls within the configured age range

Selfie Capture Types

Center only - Only one portrait of the individual facing the camera is captured
Center and profile - Three portraits are captured, with the individual facing left, center, and right of the camera

Verification Checks

See the full set of Selfie Verification checks in the Verification Checks table in your Dashboard.

Product Versions

Selfie Verifications are versioned per component. Match, Liveness, and Age are released independently, so a Selfie Verification references one version of each.

Component	Version	Notes
Match	3.3.2	Reduce false rejections caused by restricted ID portraits
Match	3.3.1	Improve reliability of account comparison results
Match	3.3.0	Expand Digital ID matching coverage
Match	3.2.0	Add support for Digital ID matching
Match	3.1.2	Reduce false rejections caused by low quality ID portraits
Match	3.1.1	Bug fix for cross product interoperability
Match	3.1.0	Improve support for NFC match
Match	3.0.0	Major version update of face comparison ensemble model
Liveness	4.3.0	Strengthen capture time defenses. Further improvements for injection attacks.
Liveness	4.2.0	Improve deepfake and generative AI recall
Liveness	4.1.0	Significant improvements for video based presentation attacks
Liveness	4.0.0	Major version update of deepfake detection ensemble model
Liveness	3.5.0	Improve generative AI recall. Minor version update of face liveness ensemble model.
Liveness	3.4.0	Improve replica detection. Minor version update of face presentation attack detection ensemble model.
Liveness	3.3.0	Significant improvements for injection attacks
Liveness	3.2.0	Significant false positive rate improvements. Adjust risk levels to account for new ensemble model score shifts.
Liveness	3.1.0	Significant false positive rate improvements. Adjust risk levels to account for new ensemble model score shifts.
Liveness	3.0.1	Significant false positive rate improvements. Adjust risk levels to account for new ensemble model score shifts.
Liveness	3.0.0	Major version update that includes improved deepfake and generative AI recall
Liveness	2.2.0	Significant false positive rate improvements. Adjust risk levels to account for new ensemble model score shifts.
Liveness	2.1.0	Improve auto-capture experience
Liveness	2.0.0	Major version update of face presentation attack detection ensemble model
Age	2.3.0	Improve detection of age inconsistencies
Age	2.2.1	Enforce stricter face quality. Exclude adverse face positioning and illuminations by default.
Age	2.2.0	Minor version update of age estimation ensemble model
Age	2.1.2	Expose additional risk configuration settings
Age	2.1.1	Improve low image quality handling
Age	2.1.0	Broader rollout of new checks: `selfie_age_threshold_comparison` and `id_age_inconsistency_detection`
Age	2.0.0	Launch new selfie age estimation model