ID Verification Overview

Document analysis and identity verification for Nigerian identity documents with graded reliability and optional regional context.

The PlatformXe ID Verification flow combines document OCR with profile name matching and optional cross-reference via Identity Resolution (NIN/BVN lookups through licensed providers). Matching is graded (FULL through NONE) with clear advisories for secondary checks — not a single legacy boolean.

Supported document types

Document	Code (hint)	Description
National Identification Number	`NIN`	NIMC NIN slip or card
Bank Verification Number	`BVN`	NIBSS BVN slip
International passport	`PASSPORT`	Nigerian passport (MRZ when present)
Driver's license	`DRIVERS_LICENSE`	FRSC license
Voter's card	`VOTERS_CARD`	INEC PVC
Tax ID	`TIN`	FIRS TIN certificate
CAC	`CAC_RC_NUMBER`	Corporate registration

Verification flow

1. Upload document → 2. OCR extraction → 3. Name matching (tiered) → 4. Optional cross-reference → 5. Result + advisories

Upload document — your app sends an image via documentUrl or documentBase64 to POST /api/v1/ocr/verify-identity.
OCR extraction — text and structured fields are extracted (Azure Computer Vision).
Name matching — profile names are compared using Nigerian-aware rules, a variation dictionary (including DB-backed entries), soft affinity priors, and optional matchingContext (marital status, state of origin code, etc.). State/LGA are never required; when provided, they act as soft disambiguators (e.g. regional naming clusters).
Cross-reference (optional) — Identity Resolution can validate identifiers against national databases (separate endpoints / scopes).
Result — you receive isVerified (strict automatic pass), nameReliabilityLevel, requiresSecondaryValidation, verificationAdvisories, and extracted fields.

isVerified: true only when name reliability is FULL and OCR confidence is ≥ 0.85. Many valid onboarding cases stop at HIGH or MEDIUM with requiresSecondaryValidation: true — design your UX accordingly (BVN, phone OTP, manual review).

Key features

Multiple document types — major Nigerian ID documents
OCR-powered extraction — with MRZ where applicable
Graded Nigerian name matching — reliability tiers, not only a single score
Optional regional context — e.g. state-of-origin code on the profile to help disambiguate similar spellings across regions
Cross-referencing — optional verification via Identity Resolution
Operational dictionary — name variations can be maintained in PostgreSQL with Redis cache

Confidence thresholds (summary)

Metric	Typical role
OCR floor for matching	~0.6 — below this, extraction is too weak for reliable name comparison
Automatic verification (`isVerified`)	`FULL` name tier + OCR ≥ 0.85
Secondary validation	Use `requiresSecondaryValidation` and `verificationAdvisories`

Poor image quality (glare, blur, crop) lowers OCR confidence and caps how high name reliability can go. Prompt users for clear, well-lit photos.

API endpoints

Document path (OCR + name match):

Method	Path	Description
POST	`/api/v1/ocr/verify-identity`	Upload image and verify names / ID text

Identity Resolution (identifier lookups):

Method	Path	Description
POST	`/api/v1/identity/verify`	Verify identity data against national databases
POST	`/api/v1/identity/resolve`	Resolve linked identifiers

Next steps

Verify Identity (OCR) — request/response reference
Identity Resolution — national database cross-reference
Verify (Identity Resolution) — direct verify API