Theme: Rollcall Automation
Problem statement: Smart attendance system using face recognition integrated
with college database
Team name: Py-Coders
Parent department: CSE
Team Members: Jerin J – CSE 3rd year
Kaviyarasan – CSE 3rd year
Mohamed Rafiq I – CSE 3rd year
Uthayakumar M D – EEE 2nd year
Team Leader: Mohamed Rafiq I – CSE 3rd Year
Date: 22/10/2025
Problem Statement
Problem: Smart attendance system using face recognition integrated with college database
The challenge: Create a secure, automated roll-call system using facial recognition that
ensures only physically present students can mark attendance.
Introduction
Our solution — SmartRoll — is a face recognition–based attendance system designed to
replace manual roll calls with a fast, accurate, and tamper-proof method.
Instead of relying on GPS or manual verification, it uses short-lived session QR codes
combined with facial recognition to confirm both the session and the student’s identity.
Approach
We merge session-based validation with biometric verification to ensure authenticity:
1. Ephemeral Session Code
o Instructor starts a class → system generates a unique QR/PIN valid for 60–90
seconds.
o Displayed on the projector or classroom board.
2. Student Verification
o Student opens web app → scans QR / enters PIN.
o App requests face capture → sends to backend for comparison with stored
face embedding.
3. Geolocation Tagging (Geo-Verification):
o The student’s device automatically captures location coordinates during
attendance marking.
o The system verifies that the student is within a predefined radius (e.g., 30–50
meters) of the classroom’s geofence.
Only when all three conditions are met does the system record attendance.
Solution Design
Teacher Module
• Create class and start session.
• Generate & display QR / PIN.
• View verified students and real-time attendance logs with geolocation data.
Student Module
• Scan QR / enter PIN to join session.
• Capture live face image for verification.
• App automatically tags geolocation and submits attendance.
• Receives confirmation once verified and logged.
Backend Process
• Validate session ID and expiration.
• Check if geolocation is within classroom radius.
• Run face recognition model and calculate confidence score.
• If all checks pass, mark attendance and store {student_id, session_id, timestamp,
coordinates, confidence}.
Workflow Architecture
1. Frontend (React / React Native)
• Handles QR scan, camera access, and UI.
• Sends { student_id, session_id, face_image } to API and checks Geolocation
2. Backend ([Link] + Express)
• Validates session and student identity.
• Generates and serves ephemeral QR codes.
• Calls Face Recognition Service for matching.
3. Face Recognition Service
• Converts face image to embeddings (using FaceNet / ML Kit).
• Compares embedding with stored reference in database.
• Returns confidence score.
4. Database (MongoDB / Firebase)
• Stores users, face embeddings, session data, attendance logs.
5. Admin / Teacher Dashboard
• Web dashboard showing real-time attendance, session stats, and analytics.
Simplified Data Flow:
Data Privacy & Security
• Store only face embeddings, not raw images.
• Encrypt data at rest and in transit (HTTPS + AES).
• Auto-delete embeddings for graduated students to ensure compliance.
Key Features
• Fast QR-based class session setup.
• Accurate facial recognition for identity.
• No GPS dependency.
• Secure, scalable cloud-ready architecture.
• Real-time dashboards for faculty.
Expected Impact
• 90% reduction in roll-call time.
• Eliminates proxy attendance.
• Provides reliable analytics on attendance patterns.
Future Enhancements
• Liveness detection (blink / movement) for anti-spoofing.
• BLE/NFC validation for precise in-room proof.
• Integration with LMS or campus ERP systems.
• Role-based analytics for departments and administration.
Conclusion:
Our solution provides a fast, secure, and intelligent approach to automating classroom
attendance through facial recognition.
By combining ephemeral session QR codes with AI-based identity verification, we
eliminate the flaws of traditional roll-call systems — ensuring that attendance is marked only
by students who are genuinely present.
The system is lightweight, scalable, and privacy-conscious, making it suitable for real-world
deployment in educational institutions of any size.
With its modular architecture and cloud-ready design, it can easily evolve to include
advanced features like liveness detection and BLE-based presence validation.
In essence, this project bridges the gap between accuracy, convenience, and security,
setting a new standard for attendance automation in the era of smart campuses.