Trust Evaluation
The IDV Toolkit provides results to evaluate trust that can be given to a remote identity verification.
When the system submitted both ID document photo and selfie photo to Digital Identity Service, it can perform the inspect functions. These return various values that can be used in “"Trust factors"”.
Based on these Trust factors, the solution using the IDV Toolkit can automatically decide if the person’s identity can be trusted or not, or if there is a human inspection and decision needed. Ideally, all decisions should be automatic. However, at the beginning of a project it is better to analyze more edge cases manually to narrow down thresholds and then reduce need for manual input.
Trust Factor | API Endpoint | Return Value | Description | Recommended Effect on the Overall Trust |
---|---|---|---|---|
Face match | customers/inspect | selfieInspection .similarityWith .documentPortrait | Similarity score between selfie and portrait on the document | Reject if false |
Passive liveness | liveness/evaluation | score | Passive liveness of the provided selfie | Reject if below threshold |
Active liveness: Smile or Eye-gaze | liveness/evaluation | score | Chosen Active liveness for the provided photos | Reject if false |
Age consistency | customers/inspect | ageDifferenceWith .dateOfBirth | Difference between age estimated for selfie and date of birth | Mark for review or reject according to difference |
Document expired | document/inspect | expired | Notes that document is expired | Reject if true |
Valid MRZ checksum | document/inspect | mrzInspection .valid | If document contains MRZ, notes that MRZ field might be modified | Mark for review if false |
Text consistency | document/inspect | textConsistency .consistent | Cross-checks the fields in visual zone with its representation in MRZ or other sources | Mark for review if false |
Screen attack detection | document/inspect | pageTampering .additionalProp1 .looksLikeScreenshot | Checks if the photo is not taken from a screen | Reject if true |