Face Verification

Match every face.Catch every imposter.

ISO-compliant 1:1 biometric face matching that compares a live selfie against an ID document photo — with 99.9% accuracy, in under 500ms, across all skin tones and lighting conditions.

Request a demo

99.9%

Match accuracy

<500ms

Response time

130+

Skin tones tested

ISO 30107

Compliant

CAPABILITIES

Not just face matching. Biometric certainty.

DigiVerify's face verification goes beyond a simple photo comparison — it analyses facial geometry, texture, and depth to deliver a mathematically grounded match decision.

1:1 Face Matching

Compares one selfie against one reference photo — typically the biometric photo extracted from an identity document. Delivers a precise similarity score from 0–100 with a configurable match threshold.

ISO 30107-3Configurable ThresholdSimilarity Score

Photo Match with Selfie

Cross-matches a live selfie against the photo on the submitted ID document. Confirms the person presenting the ID is its rightful owner — the foundational check for any KYC flow.

Exact MatchKYC OnboardingDocument + Selfie

Multi-factor Biometric Analysis

Goes beyond pixel comparison. Analyses facial geometry (landmark positions), texture patterns, depth map estimation, and micro-feature deltas to produce a robust, multi-dimensional match score.

Geometry AnalysisTexture MappingDepth Estimation

Universal Fairness

Trained and validated across 130+ skin tone variants, all age groups from 18–80, and diverse facial hair styles and accessories. Equal false reject rates across demographic groups — no bias.

130+ Skin TonesAge 18–80Bias Tested

Lighting Robustness

Maintains high accuracy across 8 tested lighting scenarios — indoor fluorescent, outdoor daylight, low-light evening, backlit, harsh flash, and mixed. Normalisation runs automatically before matching.

8 Light ConditionsAuto NormaliseLow-light Capable

Anti-Spoofing Signals

Outputs spoofing risk signals alongside the match score — flagging when the selfie shows inconsistencies with a live capture. Pair with Liveness Detection for the full anti-spoof stack.

Spoof Risk ScorePairs with LivenessDeepfake Flag

HOW IT WORKS

A match decision in four precise steps.

Face verification runs a structured biometric pipeline — from capture quality checks through to a mathematically grounded similarity score.

Selfie Capture

The user takes a selfie via your app using our pre-built SDK or your own UI. Real-time guidance checks face position, distance, lighting, blur, and angle before accepting the capture.

Face Detection & Normalisation

The AI locates, crops, and aligns the face in both the selfie and the reference image (document photo). Contrast, orientation, and resolution are normalised to a canonical form for fair comparison.

Biometric Comparison

Multi-dimensional analysis compares facial geometry (68 landmark points), texture patterns, depth map estimation, and micro-feature vectors. Each dimension contributes to the final similarity score.

Scored Decision

Returns a structured JSON result: MATCH / NO_MATCH / REVIEW decision, per-dimension scores, overall similarity score, spoof risk flag, and full audit trail — in under 500ms.

Selfie Capture

😐

Centre your face in the oval.
Look directly at the camera.

Face detected

Lighting: Good

Eyes: Open

Move closer

Face Detection & Normalisation

🪪

ID Photo

Extracted ✓

🤳

Live Selfie

Normalised ✓

Face landmarks detected (68 pts)

Alignment: Frontal pose confirmed

Resolution: Scaled to 224x224

Contrast normalised

Biometric Comparison

BIOMETRIC ANALYSIS RUNNING...

Facial geometry

98.1%

Texture analysis

97.2%

Depth map

99.6%

Landmark delta

96.8%

Feature vectors

98.4%

Processing... 248ms elapsed

Scored Decision

{
  "decision": "MATCH",
  "similarity_score": 98.4,
  "threshold_used": 85.0,
  "response_ms": 412,
  "scores": {
    "geometry": 98.1,
    "texture": 97.2,
    "depth_map": 99.6,
    "landmarks": 96.8
  },
  "spoof_risk": "LOW",
  "spoof_score": 2.1
}

USE CASES

When you need absolute certainty about who you're talking to.

Face verification is the critical check at the points in your product where impersonation risk is highest.

Customer Onboarding (KYC)

During account opening, match the user's live selfie against the photo on their submitted identity document. Confirms the person is who they claim to be — the core of any KYC flow. Pairs with document verification and liveness detection.

Account Recovery

When a user loses access to their account, a face match against their original onboarding selfie confirms identity before granting recovery access — without relying on insecure OTP-only flows.

Step-up Re-authentication

For high-value transactions or sensitive actions, trigger a face match to re-confirm the session user is still the account holder. Adds a biometric layer beyond passwords and OTPs.

Transaction Authorisation

For wire transfers, large withdrawals, or beneficiary changes, require a live selfie match to authorise. Stops account takeover fraud at the moment that matters most.

COMPLIANCE & ACCURACY

Certified biometrics. Auditable decisions.

DigiVerify's face verification is built to international biometric standards and tested to deliver fair, consistent accuracy across all demographics — not just for one market.

ISO 30107-3

Biometric presentation attack detection standard

iBeta Level 1

Certified anti-spoofing performance

Bias Audited

Equal FRR across demographic groups

Full Audit Trail

Every decision logged with score breakdown

ACCURACY BY DEMOGRAPHIC

All demographics

99.4%

Dark skin tones

99.7%

Light skin tones

99.7%

Age 18–30

99.7%

Age 50–80

93.7%

Low-light capture

91.4%

False Accept Rate0.01%

False Reject Rate0.30%

FOR DEVELOPERS

Two images in. A match decision out.

Face verification is a single REST endpoint. Pass two images — reference and selfie — and receive a scored match decision in under 500ms.

face-verify.js

// DigiVerify — Face Verification API
const result = await fetch('https://api.DigiVerify.com/v1/face/match', {
  method: 'POST',
  headers: {
    'Authorization': `Bearer ${apiKey}`,
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({

    // The two faces to compare
    reference_image: docFaceUrl, // from doc verify
    selfie_image: selfieBase64,

    // Optional configuration
    threshold: 85.0, // default, configurable
    return_scores: true, // per-dimension scores
    check_spoof_risk: true // spoof signal
  })
});

const {
  decision, // "MATCH" | "NO_MATCH"
  similarity_score, // 0–100
  scores, // per-dimension
  spoof_risk // "LOW" | "MED" | "HIGH"
} = await result.json();