DocsAPI LogoDocsAPI

Instant, Accurate Passport Verification

Read the MRZ, printed fields and face photo from ICAO 9303 passports in a single API call. MRZ check digits are validated, so a misread gets flagged instead of passing silently.

Passport with MRZ highlighted
MRZ check digits validated

Comprehensive Field Extraction

document_number
issuing_country
surname
given_names
nationality
dob
expiry_date
sex

How Accurate Is It?

We have not published a benchmark for passports yet, so we will not put a made-up number here. This is what we have measured, and how errors get caught.

97-99%

Clean Typed Text

Character accuracy on clean typed text in our 2026 OCR benchmark. Printed ID fields are typed text, but glare, wear and angled photos pull it down.

ICAO 9303

MRZ Check Digits

The machine-readable zone carries check digits for the passport number, birth date and expiry date. A misread usually fails the checksum, so it gets flagged instead of passing silently.

Fields the engine is unsure about are flagged for a reviewer instead of guessed. The best test is your own sample, so bring a few real passports captures to the demo.

Passport Extraction Workflow

1

Camera or Scan Upload

Users can capture passport via mobile camera or upload a scan through your app or web interface.

2

Auto-detect & MRZ Parse

AI detects the passport type and orientation, then reads the MRZ and validates its check digits.

3

Check Digits Validation

System validates MRZ check digits to ensure document authenticity and data integrity.

4

Optional Face ≈ Selfie Match

Compare extracted passport photo with live selfie for additional identity verification.

Time and Checks

What changes when passports are read by DocsAPI instead of typed in by hand

Time

Manual entryTyped field by field
DocsAPIAbout 1.1 s per page

Speed from our 2026 OCR benchmark, including the network round-trip. Capture quality matters more than processing time: a blurry photo costs the user a retake, so guide them to a clean, flat, well-lit shot.

Checks Before the Data Moves On

  • MRZ check digits validated against ICAO 9303
  • Field formats checked before the data moves on
  • Low-confidence fields flagged for a reviewer instead of guessed

Fraud catch rates depend on your applicant mix and capture flow, so we would rather test on your samples than quote a number.

Developer Output Example

Passport OCR Response
{
  "request_id": "pass_req_8f7d9c2e",
  "status": "success",
  "processing_time_ms": 842,
  "document": {
    "type": "passport",
    "issuing_country": "GBR",
    "document_number": "123456789",
    "surname": "SMITH",
    "given_names": "JOHN JAMES",
    "nationality": "GBR",
    "dob": "1990-01-15",
    "expiry_date": "2030-01-14",
    "sex": "M",
    "mrz_raw": "P<GBRSMITH<<JOHN<JAMES<<<<<<<<<<<<<<<<<<<<<<<\n1234567897GBR9001154M3001141<<<<<<<<<<<<<<00",
    "mrz_validity": true,
    "face_crop_base64": "[base64_encoded_image_data]",
    "confidence": {
      "document_number": 0.99,
      "surname": 0.98,
      "given_names": 0.97,
      "dob": 0.99,
      "expiry_date": 0.99
    },
    "face_match_score": 0.92
  }
}

Key Response Elements

1

document_number and expiry_date for identity verification

2

mrz_validity boolean flag for document authenticity check

3

confidence scores for each extracted field

4

face_match_score for biometric verification

Compliance & Security

ICAO 9303 Compliant

Fully compliant with International Civil Aviation Organization standards for travel document processing.

End-to-End Encryption

All passport data is encrypted in transit and at rest using AES-256 encryption standards.

Auto-Redaction

Configurable data retention policies with automatic redaction when retention period expires.

Frequently Asked Questions

Our OCR engine supports multi-script recognition including Arabic, Cyrillic, Chinese, Japanese, and Korean characters. For non-Latin scripts, we extract both the native script data and the Latin transliteration from the MRZ, ensuring accurate processing of passports from all countries.

Our AI includes advanced image preprocessing that automatically corrects for glare, blur, and poor lighting conditions. The system can enhance image quality in real-time before OCR processing, and will provide confidence scores for each extracted field so you can implement appropriate fallback workflows for low-confidence extractions.

We offer both options. On-device processing provides faster results and works offline, but with slightly lower accuracy for complex cases. Server processing gives the best accuracy and supports all passport types, but requires internet connectivity. Many clients implement a hybrid approach: on-device for speed, with server fallback for low-confidence results.

Deploy passport OCR in your app.

Start verifying passports with MRZ check digits validated on every read. Book a demo and bring a few real passport images.