Objective:

To develop a secure and intelligent online exam monitoring platform that uses AI technologies like face detection, gaze tracking, voice activity, and behavior analysis to prevent cheating during remote exams.

Key Features:

Student Panel:

• Secure login with student ID verification

• System checks for webcam and microphone availability

• Face detection and environment scan before exam start

• Live monitoring during the exam

• Warning pop-ups for suspicious behavior

• Submission of answers and auto-logout

Instructor/Examiner Panel:

• Create and schedule exams

• Set rules (e.g., full-screen mode, tab switch restrictions)

• Access live proctoring dashboard

• Review AI-generated flags and behavior reports

• Manual override to pause/terminate suspicious exams

Admin Panel:

• Manage users, exams, roles

• Monitor system performance, logs

• Export reports (violations, attendance, scores)

• Maintain list of verified instructors/students

Tech Stack:

Workflow (Step-by-Step):

1. User Authentication & Setup

• Student logs in using credentials

• Webcam/mic permissions requested

• Face is captured and matched with registered ID image using facial recognition

• Environment scanned for additional faces or devices

2. Exam Initialization

• Full-screen exam interface is launched

• System disables right-click, copy-paste, tab-switching

• Begins AI-based proctoring (video + audio feed monitoring)

3. AI Monitoring During Exam

• Face Recognition: Verifies identity remains the same

• Gaze Detection: Tracks if student is looking away frequently

• Voice Detection: Flags background conversations

• Multiple People Detection: Detects presence of others in frame

• Phone Detection (optional): Using object detection models

• AI assigns behavior score and confidence level continuously

4. Warnings and Auto-Actions

• Pop-up alerts are shown for any suspicious activity (e.g., “Please stay focused on screen”)

• Repeat violations trigger auto-termination of the exam

• All logs and video clips are stored with timestamps

5. Instructor Dashboard

• View live student video feeds (grid view)

• Get real-time flags and alerts

• Can chat with student or pause the exam

• View post-exam report (including screenshot logs & AI behavior report)

6. Post-Exam Report

• AI generates detailed report:

  • Face mismatch attempts

  • Number of tab switches

  • Time spent looking off-screen

  • Audio disturbances detected

• Instructor validates results or marks exam for review

Security Measures:

• Encrypted video/audio streams

• No data stored locally on student system

• Anti-tab switching, full-screen enforcement

• Plagiarism check on typed answers (optional)

Optional Advanced Features:

• Mobile phone detection via YOLO object detection

• Voice command detection (e.g., students asking for help)

• Biometric login (face + voice)

• Browser plugin for deeper monitoring (Chrome Extension)

• Offline detection mode with periodic uploads

Outcome:

This project ensures secure, AI-monitored remote examination for universities, competitive tests, and online certifications. It reduces human proctoring effort, increases fairness, and maintains academic integrity