Verify every travel document in seconds, not minutes.
VeriScanx is a computer-vision pipeline that inspects passports and ID documents for tampering, cross-checks passenger details against live government and international records, catches duplicate or blacklisted identities, and hands the counter officer one clear risk score — before the traveler has finished stepping up to the desk.
402183756234NRD9006014F
Manual document checks don't scale, and forged documents don't announce themselves.
Immigration counters and airport kiosks still lean on a trained eye and a UV lamp. That's slow at peak hours, inconsistent across shifts, and easy to beat with a well-made forgery, a photo-swapped passport, or a document that's already on a blacklist under a different alias.
Throughput bottleneck
Peak-hour queues grow because each document needs a manual look, even when 90%+ are genuine.
Inconsistent scrutiny
Fatigue, shift changes, and lighting mean the same forged page can pass one officer and not another.
Siloed watchlists
National blacklists, Interpol-style alerts, and prior-scan history rarely get cross-checked in real time.
One scan, six checks, one decision.
Every document that comes through the kiosk or counter scanner moves through the same pipeline, so the outcome is repeatable regardless of who's on shift.
Capture & upload
Kiosk camera, counter scanner, or mobile app ingests the document image and, where available, the NFC chip data.
OCR & MRZ parsing
Extracts the visual fields and machine-readable zone, then validates the MRZ checksum digits.
Image forensics
Error-level analysis and print-pattern checks surface splicing, re-prints, and photo substitution.
Database cross-check
Queries issuing-authority records, the national blacklist, and international watchlists in real time.
Identity resolution
Face-matches the document photo and searches prior scans for duplicate or multiple identities.
Risk scoring
A weighted, explainable 0–100 score routes the traveler to clear, secondary review, or escalation.
One decision, informed by every database that matters.
A risk score is only as good as what it's checked against. Every scan fans out to the records that could catch a fraud attempt — before the officer even looks up.
CORE
Deploys wherever the checkpoint actually is.
The same verification pipeline runs identically whether it's bolted to an airport kiosk or opened on an officer's phone at a land crossing — same six checks, same risk score, same audit trail.
Self-service kiosk
Camera-equipped kiosks at airport e-gates and terminal entries — travelers scan their own document before reaching the counter.
Counter scanner
The officer dashboard used in this demo — built for the manned counters at airports, seaports, and immigration desks.
Mobile officer app
A phone-camera version for roaming checks — land border patrols, train platforms, and event entry points with no fixed kiosk.
Batch & API access
A verification API for travel agencies, visa processors, and event organizers to screen documents in bulk ahead of time.
Built for every place a document gets checked at the border.
Airports & seaports
- E-gate & counter passport checks
- Crew & transit document screening
- Peak-hour queue relief
Land border checkpoints
- Vehicle & pedestrian crossings
- Low-connectivity offline mode
- Mobile officer scanning
Railway & metro border zones
- Cross-border rail platform checks
- High-throughput, low-dwell scanning
- Ticketing-system hand-off
Visa & consulate counters
- Pre-screening at application intake
- Duplicate-application detection
- Supporting-document cross-check
Migration & refugee centers
- Identity de-duplication at scale
- Handles damaged/worn documents
- Full audit trail for case review
Events & venue security
- Accreditation & ID verification
- Watchlist screening at entry
- Fast enough for gate throughput
Built around the ways travel documents actually get faked.
Tamper & forgery detection
Error-level analysis and font/kerning consistency flag edited regions and photo swaps.
MRZ & field validation
Checksum digits and cross-field consistency (name, DOB, document number) are verified automatically.
Blacklist & watchlist matching
Real-time lookup against national blacklists and international alert lists on every scan.
Multiple-identity detection
Searches prior scans for the same face or biographic details filed under a different name.
Real-time, explainable risk score
A weighted 0–100 score with a visible breakdown, not a black-box yes/no.
Officer dashboard & audit trail
Every scan, decision, and override is logged for review and analytics.
Six modules, one pipeline.
Each check runs independently and contributes its own weighted points to the final risk score — an officer can see exactly which module flagged what, not just a pass/fail.
Image forensics (ELA)
Error-level analysis re-compresses the image and diffs it against the original, live in the browser — no upload needed to spot a re-print.
MRZ & ICAO checksum
Real ICAO Doc 9303 checksum math validates every digit of the machine-readable zone — not a simulated pass/fail.
Blacklist & watchlist
Every document number is checked against the live blacklist table on every single scan, cross-officer and persistent.
National registry cross-check
Confirms the document number is actually registered to the name printed on it — not just that the number "looks" valid.
Face verification & liveness
Matches the document photo against a live camera capture, with a blink-based liveness check to catch a printed-photo replay.
Duplicate & mutation detection
Searches scan history for the same identity under a slightly altered name or date of birth — a common way to dodge exact-match checks.
See it catch a forgery for yourself.
Upload a document image or try a fictional specimen — the image-forensics and MRZ checks run for real, right in your browser.
Five layers, one pipeline.
What we're targeting for the pilot.
These are design targets for the prototype, not measured production figures — they're what the pipeline above is architected to hit.
Average counter processing time vs. fully manual checks
Independent checks run on every single document
Share of scans expected to route to manual review
Of scans logged with a full, auditable decision trail
Frequently asked questions.
Not a real government feed — but the public site's "Live Demo" tab does query the same live application database the officer panel itself uses, in real time, rather than a separate offline dataset. The records in it are still fictional demo data for the hackathon; wiring in an actual government/international feed is a connector swap, not a redesign, since both sides already share the same schema.
Image forensics and MRZ checksum validation run entirely client-side, so they still work offline. Checks that need a live lookup (blacklist, registry, prior-scan history) are clearly marked as unavailable rather than silently skipped, and the document is routed to manual review instead of guessing.
Yes — the score is a recommendation, not an automatic decision. Every override is logged against the officer's account alongside the original score and the reason given, so the audit trail always shows both what the system found and what the human decided.
Passports, national ID/Aadhaar-style cards, driving licences, and visas each have their own validation rules (expiry logic, minimum age, validity window) — the pipeline checks which document type it's looking at and applies the right rule set automatically.
No — it's a separate, optional check the officer can run on top of the risk score, not folded into it. It runs entirely on-device (camera never leaves the browser) with a blink-based liveness check so a printed photo held up to the camera can't pass as a live capture.
Built for SIH 2026.
See it catch a forgery, live in your browser.
No install, no account — upload a document or try a fictional specimen and watch the real tamper-detection pass run in seconds.
Run a document through the pipeline.
Upload any ID-style image, or try a fictional specimen document below. The image-forensics panel runs a real error-level-analysis pass on your image in the browser; OCR/MRZ field extraction from an arbitrary upload is still simulated, but blacklist, national-registry, and duplicate-identity checks now query the same live application database the officer panel uses, in real time — no real government systems are queried.
Drop a document image, or click to upload
JPG or PNG · processed entirely in your browser
No submit button — results update live as you type, checked against the same live blacklist, registry, and scan-history database the officer panel uses. Try any document number, or check the officer panel's Registry / Blacklist pages for what's currently on file.
Today at Terminal 2, Counter 4–9
Scan volume — last 14 days
Scans by risk band
Recent scans
| Time | Traveler | Document | Country | Risk | Status |
|---|
Identity graph THIS SESSION
Every identity (name + DOB) and document number scanned this session, and how they connect. A document linked to more than one identity, or an identity linked to more than one document, is highlighted as a conflict.
Top flag reasons — last 14 days
Identity graph — conflicts detected
Mutation detector THIS SESSION
Looks for near-duplicate identities across this session's scans that an exact-match check would miss — a spelling variant of the same name, or a date-of-birth off by a few days — a common pattern when someone submits slightly altered details across applications. Each pair gets a stability score: lower means less consistent, more likely the same person under drifted details.