AI Anti-Cheating System (ExamShield)

AI-powered online exam proctoring — real-time webcam analysis, browser lockdown, audio monitoring, violation logging, and admin review dashboards.

Features · Demo · Installation · Usage · API · Architecture · Contributing

Repository: github.com/Sathvik257/AI-Anti-Cheating-System

About

AI Anti-Cheating System is a full-stack exam integrity platform for schools and online assessments. Students take proctored exams in the browser (or optional Electron desktop app); the system detects suspicious behavior with TensorFlow.js (face + object detection), logs violations to MongoDB, and gives admins tools to review evidence and AI-generated risk summaries.

Component	Path	Default
Frontend	`AI-Exam/`	http://localhost:8080
Backend API	`AI-Exam/backend-code/`	http://localhost:3000
Database	MongoDB	Port `27018` (local) or Atlas

Features

Student experience

Feature	Description
Auth	Register, JWT login, protected exam routes
Pre-exam wizard	Rules → device checks → calibration → blocked apps → start
AI proctoring	BlazeFace (faces/gaze) + COCO-SSD (phones, books, etc.)
Audio monitor	Mic level tracking, loud-noise violations (20% threshold)
Browser lockdown	Blocks tab switch, copy/paste, DevTools, right-click, screenshots
Identity check	Face enrollment + continuous verification during exam
Exam timer	Countdown with auto-submit on critical violations
Dashboard	History, scores, violation counts at `/dashboard`
AI assistant	In-exam coach, accessibility (large text, contrast, voice hints)

Admin experience

Feature	Description
Admin dashboard	All violations & sessions — `/admin`
Analytics	Charts & trends — `/analytics`
Violation review	Mark as `benign` (false positive) or `confirmed`
AI session summary	Risk score, flags, evidence timeline (OpenAI-compatible API)
User management	List users and exams via REST API

Platform

Session video recording on submit
Multi-exam types (quiz, coding, essay, mixed)
Optional Electron shell for real OS process scanning (Teams, Zoom, Discord, etc.)

Demo

Add screenshots or a screen recording after deployment and place them here.

Screen	Description
Landing	Home page with login/register
Pre-exam setup	5-step wizard with calibration
Live exam	Webcam + violation log + timer
Admin dashboard	Violations and session review

Suggested assets (add to repo later):
  docs/screenshots/landing.png
  docs/screenshots/exam.png
  docs/screenshots/admin.png

Tech stack

Frontend     React 18 · TypeScript · Vite · Tailwind CSS · shadcn/ui
AI/ML        TensorFlow.js · BlazeFace · COCO-SSD
Backend      Node.js · Express · Mongoose · JWT · bcrypt
Database     MongoDB (local or Atlas)
Desktop      Electron 35 (optional)
Testing      Vitest · Testing Library

Installation

Prerequisites

Node.js 18+
MongoDB (local) or MongoDB Atlas
Webcam & microphone
Chrome, Edge, or Firefox (HTTPS required in production)

1. Clone the repository

git clone https://github.com/Sathvik257/AI-Anti-Cheating-System.git
cd AI-Anti-Cheating-System/AI-Exam

2. Install dependencies

# Frontend
npm install

# Backend
cd backend-code
npm install
cd ..

3. Environment files

Never commit .env files. Add them to .gitignore (already recommended).

AI-Exam/backend-code/.env

MONGODB_URI=mongodb://localhost:27018/anticheating_exam
JWT_SECRET=replace-with-a-long-random-secret
PORT=3000
JWT_EXPIRES_IN=1h
FRONTEND_URL=http://localhost:8080

# Optional — admin AI summaries
OPENAI_API_KEY=
OPENAI_BASE_URL=https://api.openai.com
OPENAI_MODEL=gpt-4o-mini

AI-Exam/.env (optional in dev)

VITE_API_URL=http://localhost:3000/api

In development, Vite proxies /api → http://127.0.0.1:3000, so you can skip VITE_API_URL.

MongoDB Atlas:

MONGODB_URI=mongodb+srv://<user>:<password>@cluster.mongodb.net/anticheating_exam

4. Start services

Terminal 1 — MongoDB (Windows)

cd AI-Exam
.\start-mongodb-27018.bat

Terminal 2 — Backend

cd AI-Exam/backend-code
npm run dev

Terminal 3 — Frontend

cd AI-Exam
npm run dev

Open http://localhost:8080

5. Create admin account

cd AI-Exam/backend-code
node create-admin.js

Field	Default
Email	`admin@example.com`
Password	`admin123`

Change these before production.

Usage

Application routes

Route	Role	Purpose
`/`	Public	Landing
`/login`	Public	Login
`/register`	Public	Student signup
`/exam/setup`	Student	Pre-exam wizard
`/exam`	Student	Live proctored exam
`/dashboard`	Student	Exam history
`/admin`	Admin	Violations & sessions
`/analytics`	Admin	Statistics & charts

Run modes

Browser only — fastest setup; blocked apps use manual student confirmation.

# MongoDB + backend + frontend (see Installation)

Electron + OS process scan — detects Teams, Zoom, Discord, Slack, AnyDesk, TeamViewer, Webex.

cd AI-Exam
npm run dev              # Terminal A
npm run electron:open    # Terminal B (or npm run electron:dev for both)

Frontend only — UI works; login/API will fail without backend.

Pre-exam flow (5 steps)

Accept integrity rules
Check internet, webcam, microphone
Calibrate lighting, face position, mic baseline
Close blocked messaging/remote apps (ProhibitedAppsGate)
Review summary → Start exam

Proctoring layers

Layer	What it does
OS (Electron)	Scans running processes before exam
Browser	Tab switch, clipboard, DevTools, fullscreen, Print Screen
Audio	Loud noise above 20%, 5s cooldown
AI vision	Face count, gaze, obscured face, prohibited objects (~2s interval)
Violations	Timestamps, optional face screenshots, admin review workflow

Auto-submit triggers (examples): multiple faces, tab switch (per exam policy).

Architecture

flowchart LR
  A[Student UI] --> B[Vite :8080]
  B -->|/api proxy| C[Express :3000]
  C --> D[(MongoDB)]
  A --> E[TensorFlow.js]
  A --> F[Electron optional]
  F --> G[OS Process Scan]

Project structure

AI-Anti-Cheating-System/
└── AI-Exam/
    ├── electron/              # Desktop shell (process scan)
    ├── backend-code/          # Express API
    │   ├── server.js
    │   ├── routes/
    │   ├── controllers/
    │   ├── models/
    │   └── middleware/
    ├── src/
    │   ├── pages/
    │   ├── components/
    │   ├── hooks/             # useAIProctoring, useWebcam, …
    │   ├── contexts/
    │   └── lib/api.ts
    ├── vite.config.ts
    └── package.json

Deep dive: AI-Exam/ARCHITECTURE.md

API reference

Base URL: http://localhost:3000/api
Auth header: Authorization: Bearer <token> (from login, stored as auth_token)

Health

GET /api/health

Authentication

Method	Endpoint	Auth
POST	`/api/auth/register`	No
POST	`/api/auth/login`	No
GET	`/api/auth/me`	Yes
POST	`/api/auth/logout`	Yes

Register example:

{
  "name": "Jane Student",
  "email": "jane@school.edu",
  "password": "securepass123",
  "studentId": "STU-001"
}

Violations

Method	Endpoint
POST	`/api/violations`
GET	`/api/violations`

Exams

Method	Endpoint
POST	`/api/exams`
GET	`/api/exams/active`
GET	`/api/exams/:id`
POST	`/api/exams/sessions/start`
POST	`/api/exams/sessions/submit`
GET	`/api/exams/sessions/history`
GET	`/api/exams/sessions/:id`

Analytics

Method	Endpoint
GET	`/api/analytics/violations`
GET	`/api/analytics/sessions`

Admin

Method	Endpoint
GET	`/api/admin/violations`
GET	`/api/admin/violations/:id`
PATCH	`/api/admin/violations/:id/review`
GET	`/api/admin/sessions`
GET	`/api/admin/sessions/:id/ai-summary`
GET	`/api/admin/users`
GET	`/api/admin/exams`

MongoDB collections

Collection	Stores
`users`	Students & admins
`exams`	Exam definitions
`examsessions`	Sessions, scores, recordings
`violations`	Events + review status

Scripts

Frontend (`AI-Exam/`)

Command	Description
`npm run dev`	Dev server → :8080
`npm run build`	Production build → `dist/`
`npm run preview`	Preview production build
`npm run lint`	ESLint
`npm test`	Vitest
`npm run electron:dev`	Vite + Electron
`npm run electron:open`	Electron only

Backend (`AI-Exam/backend-code/`)

Command	Description
`npm run dev`	Nodemon → :3000
`npm start`	Production server
`node create-admin.js`	Seed admin user

Deployment

Build frontend: cd AI-Exam && npm run build
Host dist/ behind HTTPS (required for camera/microphone)
Deploy backend with production MONGODB_URI, JWT_SECRET, FRONTEND_URL
Set VITE_API_URL to your public API URL
Rotate default admin password
Do not commit secrets; use environment variables on your host (Railway, Render, VPS, etc.)

Resource	Approximate
Bundle + models	~2–3 MB
Model load time	5–10 s
Detection interval	~2 s
RAM (with models)	100–200 MB

Troubleshooting

Issue	Fix
`Cannot connect to server`	Run `npm run dev` in `backend-code`
MongoDB error	Start `start-mongodb-27018.bat` or fix `MONGODB_URI`
CORS / Failed to fetch	Use dev proxy (`/api`) or match `FRONTEND_URL`
No admin menu	Run `create-admin.js`, re-login, check `role: "admin"`
Camera blocked	Allow permissions; use HTTPS in prod
AI models slow	Wait for first download; check network
Blocked apps stuck	Use Electron or confirm each app manually

More guides:

Contributing

Contributions are welcome.

Fork the repository
Create a branch: git checkout -b feature/your-feature
Commit changes: git commit -m "Add your feature"
Push: git push origin feature/your-feature
Open a Pull Request against main

Please do not commit .env, credentials, or API keys.

Security & privacy

Exam monitoring requires explicit student consent (rules step)
Violation screenshots are face-focused evidence for review
JWT tokens stored in localStorage — use HTTPS in production
Admin AI summaries are advisory; they do not auto-fail students
Review workflow supports false-positive marking (benign)

License

This project is private and proprietary unless the repository owner publishes an open-source license.

Author & links

GitHub: @Sathvik257
Repository: AI-Anti-Cheating-System

If this project helps you, consider giving it a star on GitHub.

Built for secure, fair online examinations with transparent admin review.

Name		Name	Last commit message	Last commit date
Latest commit History 2 Commits
AI-Exam		AI-Exam
README.md		README.md

Folders and files

Latest commit

History

Repository files navigation

AI Anti-Cheating System (ExamShield)

About

Features

Student experience

Admin experience

Platform

Demo

Tech stack

Installation

Prerequisites

1. Clone the repository

2. Install dependencies

3. Environment files

4. Start services

5. Create admin account

Usage

Application routes

Run modes

Pre-exam flow (5 steps)

Proctoring layers

Architecture

API reference

Health

Authentication

Violations

Exams

Analytics

Admin

MongoDB collections

Scripts

Frontend (AI-Exam/)

Backend (AI-Exam/backend-code/)

Deployment

Troubleshooting

Contributing

Security & privacy

License

Author & links

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Frontend (`AI-Exam/`)

Backend (`AI-Exam/backend-code/`)

Packages