Assurique Document Reader
Read any Algerian ID
In under a second
Fast capture of front and back ID images in seconds, with no manual retyping. Models trained on the Algerian national biometric ID for higher accuracy. MRZ vs visual cross-checks and format validation to catch tampering. Clear pass/fail plus reason codes for every mismatch or low‑quality field.
Live Demo
Extract ID fields in seconds
or click above to browse files
Processing
Analyzing document…
Example data: simulated extraction, not a real document
extraction
·
How does document verification work?
A fast capture flow that turns photos into clean, trusted identity data, in seconds.
Guided capture, AI extraction, and cross-checks, built for Algerian national IDs.
Benefits
One capture flow, structured identity
Guided capture, bilingual OCR on the national ID, and authenticity checks, tuned for Algerian ID cards, ready for your core systems.
Guided capture
Auto-crop, blur, and framing feedback so users get a usable image before OCR runs, fewer retries.
100% on-premise
Extraction and image analysis stay on your network, no foreign SaaS, no ID photos sent offshore.
Bilingual OCR
Arabic and Latin read in parallel with per-field confidence, models trained for Algerian documents.
Authenticity signals
Tamper, screen-replay, and edit cues surface before bad captures become trusted records.
MRZ vs visual cross-read
Compare printed zones and machine-readable lines to flag mismatches and cloned layouts.
Core-ready output
Normalized JSON with confidence and reason codes, same shape whether you auto-approve or route to review.
AI-powered OCR
Our machine learning models are trained exclusively on Algerian national ID cards, covering thousands of real documents with varying lighting conditions, wear, and capture quality variations. They understand field layouts, font variants, and position shifts specific to the Algerian biometric ID, reading Arabic and French in parallel and assigning a confidence score to every extracted field.
- Algeria-trained: purpose-built for the national biometric ID card
- Arabic & French extracted natively, including RTL name fields
- MRZ and visual zones read in parallel and cross-validated
- Per-field confidence scores: auto-approve high confidence, route the rest
Photo corrected before ID is read
Real-world captures are often off-angle, reflective, or slightly blurred. The pipeline automatically straightens and crops the document, reduces glare and motion blur, and ensures the ID is readable before extraction.
This improves OCR accuracy and makes tampering and screen replays easier to detect, reducing retakes, manual review volume, and ambiguous outcomes.
Document authenticity checks
Every scan is checked for tampering, reprinting, and forgery before any data is returned, not just whether the fields are readable.
- MRZ cross-checked against every printed field
- Security layout and check digits validated
- Print and screen-replay attempts detected
- Mismatches flagged automatically for review
OCR extraction, speed & consistency
AI-powered extraction, fast verification, and cross-checked data, everything you need for reliable national ID card identity capture.
98.7%
Field accuracy
Covering every detail we read from the Algerian national ID card
<1s
Extraction time
Full field extraction including Arabic & French, on any mobile device
15+
Fields extracted
Names, dates, places, family lines, document numbers, expiry, and more
2lang
Arabic & French
Both scripts read natively in parallel, including full RTL support
Tailored to your industry
Document Reader integrates into any onboarding flow, from regulated banking to high-volume public services.
OCR verification FAQ
Common questions about OCR-based ID capture, extraction quality, and how we handle Algerian national ID card data.
OCR is fully supported for the Algerian national ID card (front and back). For Algerian ePassports and other ICAO documents, Assurique Chip Reader provides NFC verification today. Passport photo OCR, driving licences, and other document types are on our roadmap and in development. Contact us for priorities and timelines.
Good lighting, the entire document visible in frame, and no heavy blur or glare. Our capture SDK guides the user in real time, and our OCR models are trained to handle everyday noise such as minor blur and reflections.
In typical production flows, OCR extraction and consistency checks complete in a few seconds. You can tune timeouts and fallbacks so users are not left waiting during onboarding.
Yes. Our on-device SDK performs capture and OCR locally, so the first pass works even on low or no connectivity. You can optionally send results to the server for deeper checks and archiving.
Our models are trained on both Arabic and Latin text for national ID card documents. We keep both representations when needed and normalize key fields (names, dates, numbers) so your backend receives consistent, machine-readable values.
You control where OCR runs and how long data is stored. Use ephemeral processing only, retain hashes or metadata for audit, or keep full records to meet regulatory requirements—all configurable to align with Algerian and international privacy rules.
Turn national ID cards into clean data
AI-powered OCR and national ID card validation to extract and verify identity data for onboarding and compliance.