👤 Face Attendance 🗺️ MemoMoveAI 👥 MemoVisitorAI ☕ MemoCafeAI 🧠 MemoInsightsAI 🎓 ExamIntegrityAI NEW
🎓 Module 06 · New · Exam Integrity AINEW

Exam hall integrity.
AI-verified. Zero proxy.

Purpose-built for universities and colleges. A 5-step AI validation chain verifies every student's identity, exam registration, and room assignment in real time — making proxy attendance in examination halls biometrically impossible.

≥99.5%Face match accuracy
<4sPer scan validation
10,000+Concurrent candidates
0%Proxy possible
Exam Hall C · Business Ethics Active
42 / 50
Present
3
Rejections
✓ Arjun Sharma · 0.8sAPPROVED
✓ Kavya Reddy · 1.1sAPPROVED
⛔ Rohan MehtaWrong Room
⛔ Photo Spoof AttemptSpoof Detected
✓ Priya Nair · 0.9sAPPROVED
🎓 68 minutes remaining · Admin dashboard live
The Problem

Manual exam attendance is
systematically broken

Traditional manual verification in exam halls fails on every dimension: speed, accuracy, fraud prevention, and compliance. MemoFaceAI ExamIntegrityAI is built specifically to fix this.

🎭

Proxy Attendance Is Rampant

Impersonation in exams — one student sitting for another — is a documented problem across Indian universities. Manual ID checks are easily circumvented in crowded halls.

🐢

Manual Checks Create Queues

10–15 seconds per student at entry means the first 30 minutes of every exam are consumed by entry chaos. Students miss time. Invigilators are stressed.

📁

Paper Records Are Unreliable

Attendance sheets are filled manually, transferred by hand, and compiled days after the exam. Errors and omissions are routine. Audit trails are weak.

🏛️

Room Misallocations Go Uncaught

Students sitting in the wrong hall — due to confusion or intentional fraud — are caught only when registers are reconciled. Sometimes never.

📊

No Real-time Admin Visibility

Exam controllers have zero visibility into which halls have attendance gaps, anomalies, or technical issues — until hours after the exam ends.

📋

Compliance Reports Are Manual

Generating attendance sheets, absentee lists, and anomaly reports requires 1–2 days of manual data entry. Errors are inevitable. Audit readiness is never guaranteed.

The Core Innovation

The 5-Step AI Validation Chain —
every scan, every time

All five checks execute in sequence in under 4 seconds. Every single condition must pass. Failure at any step terminates the chain with a specific, human-readable reason code.

01
🛡️ Liveness Detection
Passive anti-spoof AI verifies the scan is a live face — not a photo, screen, or 3D mask
✓ Pass → Proceed to Step 2
✗ Fail → REJECT: Spoof Detected
02
👤 Face Recognition
Match score ≥ 0.85 required against the student's enrolled 512-dimensional face embedding
✓ Pass → Proceed to Step 3
✗ Fail → REJECT: Identity Not Verified
03
📋 Exam Registration Check
Confirmed Student ID must be registered for the currently active exam session
✓ Pass → Proceed to Step 4
✗ Fail → REJECT: Not Registered for This Exam
04
🏛️ Room Allocation Validation
Student's assigned room must match the current device's room ID — no entry to wrong halls
✓ Pass → Proceed to Step 5
✗ Fail → REJECT: Not Allocated to This Room
05
🔄 Duplicate Prevention
No prior attendance record must exist for this student–exam pair in the current session
✅ ALL PASS → Attendance Marked. Welcome, [Student Name].
✗ Fail → REJECT: Already Marked Present
System Modules

Nine integrated modules —
complete exam management

📅

Exam Management

Create exams with subject, code, department, date, time, duration, and seats. Supports multiple concurrent exams. Bulk CSV upload for large schedules. Lifecycle: Draft → Published → Active → Completed → Archived.

🗂️

Student Mapping & Enrollment

Map each student to exam ID, room ID, and optional seat number. Bulk mapping via CSV. Face stored as a 512-dimensional embedding — not raw images. Auto-alerts for low-quality enrollments.

🏛️

Room Allocation Management

Create and manage exam halls with name, building, floor, and capacity. Assign students to rooms per exam. Capacity vs. count validation. Room-wise live dashboard: Present / Absent / Not Yet Scanned.

🧠

Face Recognition Engine

Deep learning 1:N face matching. ≥99.5% true acceptance rate at 0.01% false acceptance. Handles glasses, mild facial hair, varied lighting, and ±20° angle variations. Under 4 seconds on any Android and iOS or iPad device.

🛡️

Liveness Detection (ISO/IEC 30107-3)

Passive liveness — no blink or head movement required. Single-frame analysis. Detects printed photos, screen replay, and 3D masks. Liveness score returned with every scan. Failure logged with frame snapshot.

📱

Invigilator Mobile App

Native Android & iOS. Offline-capable face capture. Large, clear accept (green) / reject (red) screen optimized for hall lighting. Real-time present/total count. Exception log visible during session.

🖥️

Admin Web Dashboard

Live attendance across all exams, all halls, all departments on one screen. Drill-down to individual records. Live anomaly feed. Manual override panel with compulsory reason entry. System health monitor.

📊

Reporting & Analytics Engine

Hall-wise attendance report, absentee list, anomaly report, department-wide summary, student history report. Export formats: PDF (signature-ready), Excel, CSV for ERP integration. All instant on demand.

🔒

Audit Trail & Tamper-Proof Records

Every scan, acceptance, rejection, and manual override is permanently logged with timestamp, device ID, and actor. Records become read-only after exam completion. Digitally signed for legal validity.

User Roles

Three roles. Clear boundaries.

⚙️ System Administrator

Full access — exam controller / registrar
  • Create, configure, and manage all exam sessions
  • Upload student registrations and face enrollments
  • Configure room allocations and seating assignments
  • Real-time multi-hall attendance dashboard
  • Authorize manual overrides — all logged with reason
  • Manage invigilator accounts and permissions
  • Full system audit log access
  • Generate all post-exam compliance reports

👁️ Invigilator

Mobile app — assigned hall only
  • Operates the mobile app during exam
  • Selects assigned exam and room at session start
  • Views accept / reject feedback per student scan
  • Receives exception alerts in real time
  • Sees present count vs. hall capacity live
  • Cannot modify student data or exam configuration
  • Cannot override rejection decisions
  • Limited to explicitly assigned hall(s) only

🎓 Student

Passive role — face capture only
  • No active system login required
  • Pre-enrolled once — face linked to Student ID
  • Presents face at hall entry for scanning
  • System checks registration, room, and liveness
  • Receives verbal/visual feedback from invigilator
  • Cannot be marked present in wrong room
  • Cannot be marked present twice in same exam
  • Face embedding stored — not raw photo
Rejection Logic

Every rejection is specific, human-readable, and actionable

All rejection messages are non-technical, unambiguous, and tell the invigilator exactly what to do next.

Error CodeMessage Shown on DeviceInvigilator Action
ERR_LIVENESSSpoof Detected. Please present live face.Ask student to face camera directly; retry once. Do not allow entry.
ERR_FACE_MATCHIdentity Not Verified. Face not recognized.Escalate to admin immediately. Do not allow entry under any circumstance.
ERR_NOT_REGISTEREDNot Registered for This Exam. Check exam schedule.Check student's hall ticket. Escalate to exam controller.
ERR_WRONG_ROOMNot Allocated to This Room. Check your hall ticket.Guide student to their correct hall. Do not admit to this hall.
ERR_DUPLICATEAlready Marked Present in This Exam.Flag for investigation. Do not re-mark. Escalate to admin.
SUCCESS✅ Attendance Marked. Welcome, [Student Name].Student proceeds to their assigned seat.
Exam Day Workflow

Four phases — fully orchestrated

Phase 1 · T-24 Hours
Pre-Exam Setup (Admin)
1Admin creates exam session in dashboard — subject, date, time, departments, room assignments
2Admin uploads student-to-exam and student-to-room mappings via CSV. System validates for missing enrollments, duplicates, and capacity overflows
3Admin creates invigilator accounts and assigns each to specific exam + room
4System pre-loads face embeddings into edge cache on each assigned device (if offline mode enabled)
5Admin publishes exam — status changes from Draft to Published. Invigilators can now see it in their app
Phase 2 · T-0
Exam Hall Entry (Invigilator)
6Invigilator opens MemoFaceAI app, selects assigned exam and room, authenticates with credentials
7Invigilator taps 'Start Attendance Session' — app confirms exam is Active in the cloud
8Student approaches with hall ticket. Invigilator points device camera at student's face
95-step validation chain executes: Liveness → Face Match → Exam Check → Room Check → Duplicate Check
10Result shown within 1.5 seconds — green screen (accept) or red screen (reject with specific reason)
11Accepted: attendance marked in cloud, student proceeds to seat. Rejected: invigilator takes action per reason code
Phase 3 · During Exam
Live Monitoring (Admin)
12Admin dashboard shows live attendance counts per hall, updating in real time as students are scanned
13Anomaly alerts appear for: liveness failures, room mismatches, duplicate attempts
14Admin can view current scan rate and intervene if technical issues arise in any hall
15Manual override available for exceptional cases — compulsory reason entry logged in permanent audit trail
Phase 4 · Post-Exam
Reporting & Archival (Admin)
16Invigilator ends session — all local data syncs to cloud with confirmation shown on app
17Admin triggers report generation: attendance sheet, absentee list, anomaly summary — instant, not days
18Reports auto-exported to configured email IDs and/or ERP endpoint in PDF, Excel, and CSV
19Exam status changes to Completed — records become read-only and tamper-evident. Permanent audit trail sealed
Why Switch

Manual exam attendance vs
ExamIntegrityAI

MetricManual / PaperExamIntegrityAI ✦
Proxy attendanceUnquantified / High Risk✓ 0% — Biometrically Impossible
Hall entry validation time10–15 sec per student✓ <4 seconds per student
Room mismatch detection✗ Not caught in real time✓ Rejected instantly at entry
Duplicate prevention✗ Manual check (error-prone)✓ Auto-blocked — 100% reliable
Attendance data availabilityHours after exam / manual entry✓ Real-time, during exam
Admin visibility during exam✗ Zero✓ Live multi-hall dashboard
Report generation time1–2 days of manual work✓ Instant on demand
Audit trailPaper (losable, alterable)✓ Tamper-proof, digital, permanent
Face match accuracyHuman — variable, fatiguable✓ ≥99.5% — consistent every scan
Data Architecture

Designed for universities at scale

Every entity in the system is designed to handle 10,000+ concurrent candidates across unlimited departments and exam halls with no performance degradation.

Student
  • student_id (UUID)
  • name, enrollment_no
  • department, photo_url
  • face_embedding (512-dim)
  • enrollment_date, status
Exam
  • exam_id (UUID)
  • subject_name, subject_code
  • department_id, date
  • start_time, end_time
  • status (Draft→Archived)
ExamRoom
  • room_id (UUID)
  • room_name, building
  • floor, capacity
  • exam_id (FK)
StudentExamMapping
  • mapping_id
  • student_id (FK)
  • exam_id (FK)
  • room_id (FK)
  • seat_no (optional)
AttendanceRecord
  • record_id (UUID)
  • student_id, exam_id
  • match_score, liveness_score
  • device_id, timestamp
  • status
RejectionLog
  • log_id
  • student_id, exam_id
  • error_code, timestamp
  • device_id
  • frame_hash (audit)
InvigilatorSession
  • session_id
  • invigilator_id, exam_id
  • room_id, device_id
  • start_time, end_time
  • sync_status
AuditLog
  • audit_id
  • actor_id, action_type
  • target_entity, target_id
  • reason, timestamp
  • ip_address

The right student. The right exam.
The right room. AI-verified.

MemoFaceAI ExamIntegrityAI is purpose-built for universities. Request a demo configured for your institution's exam scale and structure.