Campus Netra
AI-assisted campus facility management, with a live digital twin, IoT health monitoring and Lost & Found.
Report a campus problem with a photo and a location. Campus Netra classifies it, routes it to the right team, opens a work order, and shows the affected equipment changing colour on a live map of the campus until it is fixed.
🌐 Web: campusnetra.dpdns.org · 📱 Android: app available (APK)
Features
| Area | What it does |
|---|
| Issue reporting | Report a problem with photos and an exact location (building, floor, room, equipment). AI suggests the category and priority. |
| QR reporting | Every asset can carry a QR sticker. Scanning it opens a report with the location already filled in. |
| Smart routing | Issues go to the right department and technician by specialisation and current workload. Likely duplicates are flagged. |
| Work orders & SLA | Assignment, progress, parts and costs, verification and closure, each with response and resolution targets. |
| Digital twin | A live 3D and floor-plan view of the campus. Equipment turns red, blue, amber, purple or green as its condition changes, in real time. |
| Event replay | See how the whole campus looked at any moment in the past. |
| Inspections | Checklist-based inspections. A failed critical check raises a complaint automatically. |
| IoT health monitoring | Sensor devices report power, fan and light status, temperature and humidity. Confirmed faults schedule an inspection and alert staff. A history chart shows the readings over time. |
| Lost & Found | Lost and found reports are matched automatically on photo, description, place, category and time, with a conservative threshold before anyone is notified. |
| Predictive maintenance | Ranks equipment by failure risk using an explainable, weighted model. |
| Analytics & simulation | Hotspots, repeat failures, team performance, cost, and "what-if" surge simulation. |
| AI assistant | Answers questions about the platform and the user's own data, and can file reports on their behalf. |
| Mobile | Phone-friendly layout with bottom navigation, plus an Android app. |
How it works
Report (web, app or QR)
│
▼
AI classification ──► department & technician routing
│
▼
Work order ──► repair ──► verification ──► closed
│
▼
Digital twin updates live: 🔴 fault → 🔵 in repair → 🟢 healthy
IoT devices feed the same loop:
Sensor reading ──► fault confirmed ──► inspection scheduled ──► technician checks
│
no fault ◄─────────────────────┼────► fault confirmed
(back to 🟢) complaint + work order
A piece of equipment returns to green only when nothing else is still open against it.
Roles
| Role | Typical use |
|---|
| Student / Teacher | Report issues, track complaints, Lost & Found, campus map |
| Technician | Work orders, inspections, assets, digital twin |
| Facility Manager | Live issues, assignment, analytics, simulation |
| Administrator | Users and roles, campus setup, configuration, IoT devices, audit |
Access is permission-based: an administrator decides what each role can see and do.
Screenshots
Real screens from the live app. Personal details (names, emails, phone numbers, ID numbers) are blurred. Click a section to open it, and any picture to see it full size.
Open to everyone: the website, sign-in and account creation.
What a student (or teacher) sees: reporting problems, following them up and Lost & Found.
What a technician sees: assigned work, inspections and equipment.
What an administrator sees: the whole campus, people, configuration and IoT. Names and contact details are blurred.
Tech stack
- Frontend: React, Vite, Tailwind CSS, Recharts, Three.js / MapLibre for the 3D and outdoor maps
- Backend: Python, FastAPI, async SQLAlchemy, WebSockets
- Database: PostgreSQL, plus a time-series store for sensor history
- IoT: ESP32 devices over MQTT
- AI: language models for classification, matching and the assistant, each with a rule-based fallback, so the platform works without them
- Mobile: Capacitor (Android)
Project structure
CampusNetra/
├── frontend/ Web app (and the Android project in frontend/android)
├── backend/ API, background jobs and AI services
├── database/ Schema migrations and seed data
├── docs/ Diagrams (DFD, class, sequence, state, use case)
└── scripts/ Development helpers
Running locally
Requirements: Node.js 20+, Python 3.11+, PostgreSQL 16.
Backend
cd backend
python -m venv .venv
.venv/bin/pip install -r requirements.txt # Windows: .venv\Scripts\pip
cp .env.example .env # then fill in your own values
.venv/bin/uvicorn app.main:app --reload
Frontend
cd frontend
npm install
npm run dev
Android app (requires the Android SDK and JDK 21)
cd frontend
npm run build:app
cd android && ./gradlew assembleDebug
Configuration lives in environment variables (backend/.env). Never commit real credentials. .env.example lists what is needed, without values.
Security
- Role and permission checks on every request
- Short-lived access tokens with single-use refresh tokens, and one active session per account
- Built-in image captcha, login lockout and rate limiting on code and SMS requests
- AI assistant guarded against revealing credentials or other users' data
- Secrets are supplied through the environment, never stored in the code
© Campus Netra. All rights reserved.
