How Document Verification Works
Verification flow, document requirements, and what AdvanGuard checks during document verification.
This page describes the verification flow and basic document requirements.
Verification flow
A full document verification consists of four stages:
1. Document submission
Users capture or upload documents through the SDK.
- Supports live capture and file upload
- Documents with data on both sides must be submitted front and back as prompted
2. Recognition and data extraction
The system identifies the document's country/region and type (for example, a national ID, passport, or driver's license) and:
- Confirms the document is within your supported scope
- Checks whether photos are clear and complete
- Extracts fields such as name, document number, date of birth, and expiry date
Tip: Once extraction completes, you can retrieve OCR results to pre-fill forms or show progress without waiting for the full verification chain to finish.
3. Authenticity and compliance checks
The system further analyzes images and extracted data, including:
- Authenticity: Signs of photocopy, scan, screenshot, recapture, or tampering
- Integrity: Layout and fields consistent with official document formats
- Compliance: Age, validity period, nationality, and other rules you configure
Multiple signals are evaluated together; a single indicator alone rarely drives the final decision.
4. Verification outcome
When verification completes, you receive a clear GREEN (pass) or RED (fail) result.
- Failures include rejection reason categories (for example, image quality, unsupported document, suspected forgery, or expired document)
- Detailed results and extracted fields are available in the console
Document requirements
The document itself
- Must be within its validity period (rules configurable in the console)
- Free of heavy damage, tears, or obscured key information
- Name, date of birth, document number, and other key fields must be readable
- All documents submitted for one applicant must belong to the same person
Document contents
Where applicable, the document should include:
- Holder's name, date of birth, and portrait photo
- Document number
- Issue date or validity period
Photo requirements
- Formats: JPG, JPEG, PNG
- Color images; original capture or compliant scan; not edited with image software
- All four corners visible; no fingers or objects blocking the document
- Two-sided documents must be uploaded as separate front and back images
Supported languages
The system recognizes many scripts—including Chinese, Japanese, Korean, Cyrillic, and Arabic—and extracts both local and Latin characters where possible.
- Most languages support transliteration or translation into English (configurable per Level)
- For some languages, if the document contains only local script, English names may require additional processing; see actual extraction results
What we check
Image quality
Blur, glare, underexposure, overexposure, occlusion, incomplete framing, and similar issues guide users to recapture or flag items for review.
Document authenticity and integrity
Image provenance, template layout, fonts, and design consistency help detect photocopies, scans, screenshots, recaptures, tampering, and synthetic generation.
The presence or absence of a single security feature alone does not reliably indicate authenticity; the system evaluates documents holistically.
Data and business rules
Extracted fields are checked against rules you configure, such as:
- Applicant age within allowed bounds
- Remaining document validity meeting your threshold
- Nationality not on restricted lists
Updated about 2 months ago