Project Proposal: AI-Powered Examination System with Built-in Integrity Monitoring
1. Introduction
With the increasing adoption of online education, ensuring academic integrity in online
examinations has become a significant challenge. Existing proctoring tools are expensive,
require external software, and do not fully prevent misconduct. This project proposes an AI-
powered examination system with built-in integrity monitoring, integrating AI-based
proctoring, real-time monitoring, and security features to ensure fair and reliable
assessments.
2. Problem Statement
Online examinations often suffer from:
Lack of secure monitoring, leading to increased cheating.
The need for third-party proctoring software, which is costly.
The inability to track student behaviour (face movement, gaze tracking, or external
voices).
Web-based exams being easier to bypass through multiple browser instances or
VPNs.
This project aims to create a self-contained application that enhances exam security through
AI-driven proctoring and monitoring.
3. Objectives
Develop a standalone examination system with an integrated AI-based integrity
monitoring system.
Utilize real-time face tracking, gaze detection, and voice analysis to detect
suspicious behaviour.
Implement screen-locking & restricted access to prevent unauthorized activities.
Ensure encrypted data storage & automatic anomaly detection reports for
examiners.
4. Why an Application Instead of a Web-Based System?
Feature Application (Proposed) Web-Based System
Security High - Runs in a controlled Lower - Can be bypassed using VPNs,
environment with restricted access. multiple browsers, or extensions.
AI Integration Better - Can leverage GPU Limited - AI models in web browsers
acceleration for real-time AI analysis. are less efficient.
System Resource Full access - Can lock screens, disable Limited - Cannot fully prevent
Access external apps. students from switching apps.
Offline Capability Can run exams offline & sync results Requires an active internet
later. connection.
Scalability Moderate - Requires installation but High - Accessible from any device but
offers better control. weaker security.
Conclusion: A desktop application is the superior choice as it allows for better AI
integration, tighter security, and offline capability compared to a web-based system.
5. Proposed System Features
Feature Description
Secure Exam Portal Time-based exams with auto-save & encryption.
AI Proctoring Detects multiple faces, gaze tracking, and head movements.
Voice Monitoring Identifies background voices or external communication.
Screen Locking Prevents switching apps, taking screenshots, or using browsers.
Exam Integrity AI assigns a cheating probability score based on detected
Score anomalies.
Report Dashboard Teachers review real-time alerts and integrity reports.
6. Technology Stack
Component Technology
Programming Python (for AI & UI)
Language
UI Framework PyQt / Tkinter
AI Models OpenCV, YOLO, DeepSpeech (for voice detection)
Database SQLite / PostgreSQL
Security Data encryption, session tracking, and restricted system access
7. System Architecture
User Interface Python(Tkinter), MySQL/PostgreSQL, Session Tracking
(Student Exam Portal)
Exam Management (Question Handling, Timer, Submissions)
Python(Tkinter), MySQL/PostgreSQL, Session Tracking
AI-Based Proctoring
(Face Tracking, Eye Movement, Voice)
Yolo, OpenCV, DeepSpeech
Security & logging
(Screen Lock, Data Encryption, Report)
Report, Marks, Screen Lock, Data Encryption
8. Conclusion
This project offers an innovative and secure approach to online examinations, ensuring
fairness and reliability through AI-based proctoring, advanced monitoring, and built-in
security measures. With a desktop application, we can fully utilize AI acceleration via
GPU and prevent bypassing methods commonly used in web-based systems.
This proposal is scalable, cost-effective, and highly applicable to modern education
institutions.
