ExtractFox

Extract data from ID documents

Read every printed field off an ID — driver's license, national ID card, residence permit — and get them back as clean JSON, CSV, or Excel.

Drop a PDF or image here, or browse
PDF or image · up to 20 MB
Processed in-flight — never stored on our servers.

What should we pull from this id document?

Or pick specific fields

Or describe it yourself

Why this matters

Onboarding flows have to handle every ID format their users carry. Building per-country parsers is a losing battle. ExtractFox uses a vision model that reads identity documents generically, returning a consistent schema across hundreds of formats. A fintech onboarding flow that only parses US driver's licenses loses conversions when a German user uploads a Personalausweis — support tickets pile up asking for manual review. Hotel front desks photograph IDs in dim lobby lighting; traditional OCR reads 'DOB' as '08' instead of 'O8' and rejects a guest at 11pm.

How it works

  1. Step 1
    Upload the ID image

    Front (and optionally back) of a driver's license, national ID, or residence permit. Photo or scan.

  2. Step 2
    Extract identity fields

    Names, document number, date of birth, sex, dates of issue/expiry, issuing authority, address (where printed), and any machine-readable code.

  3. Step 3
    Push to your KYC pipeline

    JSON straight into your verification flow, or Excel for a manual-review queue.

Common use cases

KYC onboarding — verify identity at signup
Hospitality check-in — read guest ID at the front desk
Age verification — extract DOB to confirm legal age
Driver verification — fleet, rideshare, delivery operations
Visa and immigration intake — residence-permit fields for downstream forms
Bank account opening — KYC document capture
Rideshare onboarding — driver's license class, endorsements, and expiry for fleet compliance
Tenant screening — national ID or residence permit fields for rental application intake

Sample output

Example: US driver's license

document_typeDriver License
issuing_countryUS
issuing_stateCalifornia
document_numberX1234567
surnameDOE
given_namesJANE
date_of_birth1990-04-12
sexF
address123 Main St, San Francisco, CA 94103
date_of_issue2024-03-01
date_of_expiry2029-04-12
issuing_authorityCalifornia DMV
endorsements
restrictions
  • Corrective lenses

Frequently asked questions

How do I extract data from a driver's license or national ID?+

Upload a photo or scan of the front (and back, if it has data). ExtractFox returns names, document number, date of birth, dates of issue and expiry, address, and any other printed fields.

Does it work for IDs from any country?+

Yes. The model recognizes ID structure across hundreds of national formats. EU national IDs, US state driver's licenses, UK provisional licenses, India Aadhaar (visible fields only), and most APAC and LATAM IDs work.

Can it read the back of an ID with the barcode or MRZ?+

It extracts visible printed fields including the MRZ if present. Barcode decoding (PDF417 on US driver's licenses) is best done with a barcode reader; ExtractFox returns the visual fields.

How is this different from the passport extractor?+

Passports follow ICAO's standard and parse with a passport-specific schema (including both MRZ lines). The ID document extractor handles non-passport IDs that don't follow ICAO — driver's licenses, national IDs, residence permits.

Is it safe to upload IDs?+

Files are processed by ExtractFox's secure extraction engine and are not stored by ExtractFox. For regulated KYC workflows, contact us about a self-hosted or VPC deployment.

Can I detect tampering or fraud from the extracted data?+

ExtractFox extracts data, not authenticity signals. Pair extraction with a dedicated identity-verification provider (Onfido, Veriff, Persona) for tamper detection.

Does it work on photos taken with flash glare on laminate IDs?+

Mostly. Moderate glare and reflection on laminated cards are handled — the model reads around hotspots the way a human would. Severely overexposed or fully obscured fields may need a retake.

Can I extract both front and back of a driver's license in one request?+

Yes. Upload a single image with both sides visible, or upload front and back as separate images in one session. Endorsements and restrictions on the back are captured alongside front-side identity fields.

Compare to alternatives

ExtractFox vs Azure Document Intelligence
Azure Document Intelligence (formerly Form Recognizer) is powerful inside the Azure ecosystem and overkill outside it. ExtractFox covers the same document types — invoices, receipts, IDs, contracts, layout — without an Azure subscription, resource group, or model-deployment workflow.
ExtractFox vs AWS Textract
AWS Textract returns OCR blocks and rectangles; you write the code that turns those into the fields you actually want. ExtractFox returns the fields directly — vendor, totals, line items, parties — with no post-processing layer.
ExtractFox vs Mindee
Mindee has consolidated into one platform where every document type — invoices, receipts, IDs — draws from the same page-credit pool, priced per page processed. ExtractFox is also one endpoint for every document type, but priced as a flat monthly extraction count instead of metering by page, and it comes with a UI for people who don't want to touch the API.

Related extractors

Extract data from a passport
Upload a passport photo or scan and extract MRZ, name, passport number, nationality and expiry as JSON or Excel. Free, no signup.
Extract data from filled forms
Extract field values from any filled form — handwritten or printed. Intake forms, applications, surveys, government forms — to JSON or Excel.
Extract data from images
Image data extraction: pull structured fields from photos and screenshots — receipts, tables, IDs, charts — into Excel or JSON. Not plain OCR. Free, no signup.

From the blog

WorkflowJune 28, 2026
How to extract data from an ID document (3 methods)
Compare Tesseract OCR, MRZ parsing (TD1/TD2), and AI to pull identity fields from driver's licenses, national IDs, and residence permits — any country.