Remote E-Proctoring System
Built a comprehensive remote proctoring system employing multiple machine learning models to assist administrators in detecting cheating during large-scale exams.
Project Overview
Developed an advanced e-proctoring solution as my undergraduate final year project to address the growing need for remote exam monitoring. The system features five continuous monitoring components: facial recognition using dlib for user verification, audio analysis with speech recognition and NLP for detecting suspicious conversations, gaze detection to monitor attention patterns, person counting to ensure single-user presence, and object detection to identify unauthorized devices like phones. The backend is built with Python using TensorFlow, OpenCV, and various ML libraries, while the frontend uses React for a user-friendly interface. The system integrates with AWS S3 for secure cloud storage and Firebase for real-time data management. This project provided comprehensive experience in computer vision, machine learning model integration, cloud architecture, and building scalable web applications for educational technology.
Key Features
- ✓User verification through face recognition
- ✓Audio analysis using NLP techniques
- ✓Gaze detection and tracking
- ✓Person counting in exam environment
- ✓Object and phone detection
- ✓Real-time cheating behavior identification
- ✓AWS S3 cloud storage integration
- ✓Firebase real-time database
- ✓Administrative monitoring dashboard
- ✓Multi-user exam session support
Technical Challenges
- ⚡Real-time processing of multiple ML models
- ⚡Accurate face recognition under varying conditions
- ⚡Audio analysis for suspicious conversations
- ⚡Balancing security with user privacy
- ⚡Scalable cloud infrastructure design
Technologies Used
Project Info
Collaboration
University
Team
Lead Developer & Final Year Student
Screenshots
