Lightweight • Real-Time • Privacy-Preserving

EdgeAI-Based Online Exam Proctoring System

This research presents a scalable and privacy-preserving online exam monitoring system that detects suspicious activities in real time using lightweight AI models executed locally on student devices.

Explore Project

Core Focus

Real-time fraud detection without cloud dependency.

Domain

This research focuses on designing a lightweight Edge-AI framework to ensure secure, private, and low-latency online examinations.

Literature Survey

Existing proctoring systems rely heavily on cloud-based processing.

Research Gap

Lack of efficient real-time local processing and privacy-preserving solutions.

Research Problem

Ensuring secure real-time monitoring while minimizing cloud dependency.

Research Objectives

Develop a lightweight, privacy-preserving Edge-AI system.

Methodology

Combining object detection, behavior tracking, audio and biometric analysis.

Technologies

TensorFlow Lite, YOLO, MobileNet, FaceNet, TinyML, OpenCV.

Milestones

Project Proposal

Initial concept and methodology presentation.

Progress Presentation 1

Early development and findings.

Progress Presentation 2

System implementation and integration.

Final Assessment

Final system and documentation evaluation.

Documents

Access project documents via cloud storage.

Project Charter Proposal Document Checklist Final Reports

Presentation Slides

Access presentation slides via cloud storage.

Proposal Progress 1 Progress 2 Final

About Us

We are final-year undergraduate students from SLIIT specializing in Data Science, conducting research on Edge AI-based exam monitoring systems.

Research Team

Malaviarachchi M.A.T.T

Malaviarachchi M.A.T.T

📷 Object Violation Detection

IT22551634

Detects prohibited items such as phones, books, notes, and unauthorized persons through real-time webcam-based object detection on edge devices.

Weerasinghe R.G.H.I

Weerasinghe R.G.H.I

👁️ Behavioral Anomaly Tracking

IT22559586

Tracks head movement, gaze direction, and unusual behavioral patterns to identify suspicious student activity during online examinations.

Wijesinghe W.A.H.U

Wijesinghe W.A.H.U

🎤 Audio Integrity Verification

IT22320100

Monitors microphone input to detect background voices, second-person speech, and other audio-based exam integrity violations in real time.

Wijerathne M.M.G.S.A.B

Wijerathne M.M.G.S.A.B

🧠 Continuous Biometric Authentication

IT22212818

Continuously verifies the student’s identity using facial biometric features processed locally to maintain exam integrity throughout the session.

Supervisors

Dr. Dharshana Kasthurirathna

Dr. Dharshana Kasthurirathna

Assistant Professor - SLIIT

Machine Learning • Network Science • Distributed Systems

Sanjeevi Chandrasiri

Sanjeevi Chandrasiri

Senior Lecturer - SLIIT

AI • Neural Networks • NLP • Medical Imaging

Contact Us

Thenuri: it22551634@my.sliit.lk

Hashan: it22559586@my.sliit.lk

Heshani: it22320100@my.sliit.lk

Arunabhaya: it22212818@my.sliit.lk

Institution: SLIIT