AgentHack submission type
Enterprise Agents
Name
Kritika Yadav
Team name
Al-KNights
Team members
@Kritika_Yadav, @Nitika_Kashyap
How many agents do you use
One agent
Industry category in which use case would best fit in (Select up to 2 industries)
Universities/Academy
Complexity level
Intermediate
Summary (abstract)
EduGuard is an AI-powered smart proctoring agent built using UiPath and Python that ensures fairness in online exams. It monitors student activity in real time by detecting faces via webcam, capturing screen activity, tracking tab switches, and analyzing suspicious behavior. Without requiring constant human supervision, EduGuard automatically generates reports and sends alerts, acting like a virtual invigilator and helping institutions scale secure, cheat-free digital exams.
Detailed problem statement
With the growing shift toward online education and remote assessments, schools and universities are increasingly using platforms like Google Forms or LMS portals to conduct exams. However, these systems lack proper invigilation tools, making it easy for students to cheat by switching browser tabs, using mobile phones, or seeking external help during the exam.
Most educational institutions do not have access to advanced proctoring tools or real-time monitoring systems. Manual methods like asking students to turn on their webcams or reviewing recordings after the exam are ineffective, time-consuming, and not scalable. Human invigilators cannot monitor dozens or hundreds of students simultaneously, especially when exams are conducted online.
There is currently no integrated, intelligent, and affordable solution that can automatically monitor student behavior, detect suspicious activities, and alert faculty in real-time — without constant human oversight. This results in compromised exam integrity and unfair grading.
Detailed solution
To address the challenges of online exam monitoring, we developed EduGuard — a smart, AI-driven exam proctoring agent that combines the automation power of UiPath with Python-based behavioral intelligence. EduGuard simulates the actions of a real human invigilator by monitoring the webcam, tracking screen and tab activities, detecting suspicious behavior, and automatically reporting violations in real time.
Our solution is designed around six intelligent agents, each implemented as a UiPath process or automation component:
1. Webcam Monitor (Python + UiPath Integration)
Detects whether a student is visible during the exam using face detection via OpenCV.
Flags absence of face or multiple faces as suspicious.
On trigger, UiPath captures the timestamp and passes the event to the report generator.
2. Screen Capture Agent
Uses Python and UiPath to take screenshots only when a suspicious activity is detected (e.g., tab switching or multiple faces).
Screenshots are stored locally and referenced in the final report.
3. Tab Switch Detector (Windows API + UiPath trigger)
Monitors active window titles using Python win32gui module.
If a student switches from the exam window to another application (e.g., browser, notepad), it triggers a screenshot and logs the event.
4. Suspicious Behavior Analyzer
This agent runs rule-based analysis (e.g., 3 violations within 5 minutes).
Optionally integrates machine learning for more advanced pattern detection.
If the behavior is deemed highly suspicious, it signals the Email Sender agent.
5. Report Generator
Collects all event logs (face absence, tab switch, multiple faces) and generates a report.csv file.
Optionally formats the data into a professional PDF summary using UiPath’s Excel and Word activities.
6. Email Sender Agent
Automatically sends the report and any evidence screenshots to the invigilator/faculty email via UiPath’s SMTP or Outlook activities.
Subject lines and body content are dynamically generated based on the event severity.
Integration & Flow:
The system is built in a hybrid model:
Python scripts handle real-time data collection, detection, and triggering suspicious events.
UiPath handles automation workflows: starting Python scripts, monitoring logs, generating reports, sending alerts.
All components communicate via shared folders, logs, and file watchers in UiPath.
Security & Privacy:
All webcam/screen data is processed locally to preserve student privacy.
Only evidence of suspicious behavior is stored or sent for review.
This solution ensures scalable, cost-effective, and intelligent exam supervision — minimizing manual effort while maintaining academic integrity.
Demo Video
Expected impact of this automation
EduGuard significantly enhances the integrity and efficiency of online exams by automating the entire proctoring process. With this solution in place:
80% reduction in manual invigilation time: Faculty no longer need to monitor students live or review hours of webcam footage manually.
100% real-time detection of suspicious activity: Tab switches, missing faces, and multiple face detections are instantly flagged and logged.
Automated report generation: Proctoring reports are compiled automatically and emailed with evidence, saving hours of administrative work.
Reduces cheating probability by over 60%: The presence of a smart surveillance agent acts as a deterrent for students considering malpractice.
Scalable to thousands of students: Institutions can deploy the solution across multiple systems without needing additional staff.
Low-cost and resource-light: Built using UiPath Community Edition and open-source Python tools, making it affordable for schools and universities.
This automation bridges the gap between digital learning and secure assessments, empowering educators to maintain fairness and credibility in remote exams.
UiPath products used (select up to 4 items)
UiPath Studio
Automation Applications
Not applicable (custom exam monitoring application)
Integration with external technologies
Python, OpenCV, PyAutoGUI, win32gui, TensorFlow (optional for AI analysis), Gmail SMTP (for email alerts)
Agentic solution architecture (file size up to 4 MB)
![Screenshot_2025-07-07_220130[1].png](https://global.discourse-cdn.com/uipath/original/4X/6/9/b/69b9ab9c371aae1bf2cd9201929fa85486e96973.png)
Sample inputs and outputs for solution execution
Inputs:
Webcam feed – Captures real-time video to detect face presence or multiple faces.
Screen content – Monitors and captures screenshots when suspicious activity is detected.
Active window title – Used to detect tab switching or window changes.
Exam configuration (optional) – Exam window name, duration, and student ID (can be passed as variables).
Outputs:
report.csv – Log file of all suspicious events with timestamp and reason (e.g., “No face detected at 10:32 AM”).
Screenshots – Saved images for each suspicious event (e.g., tab switch, multiple faces), named with timestamp and reason.
Email alert – Sent to the invigilator containing summary of suspicious behavior and attached evidence.
Summary PDF (optional) – Auto-generated PDF report compiled from CSV and screenshots.