🦺 SafeSight – AI Powered Industrial Safety Monitoring System

Hackathon Project | WIET Hackverse 2.0 | Team: Stranger Strings

---

🚨 Problem Statement

Industrial worksites frequently suffer accidents due to failure in PPE compliance, unauthorized zone access, and lack of real-time worker monitoring. SafeSight addresses this using computer vision and AI to:

• Detect PPE compliance (helmet, boots, vest, goggles)
• Manage worker attendance via QR code scanning
• Monitor restricted zones and alert supervisors in real time

---

👥 Team Members & Roles

Member	Role	Responsibilities
Tanaya	Product & Pitch Lead	Prepare presentation, train models, collect PPE datasets
Varun	Backend Developer	Build API, connect MongoDB, manage authentication
Atharva	AI Engineer	Integrate PPE detection and run real-time monitoring
Archit	Frontend Developer	Build dashboard UI and camera grid interface

---

🛠️ Tech Stack

Frontend

• ⚛️ React – UI framework
• 🎨 Tailwind CSS – Utility-first styling
• 🌐 React Router – Client-side routing
• 📊 Recharts – Dashboard charts and visualization
• 🧭 Lucide React – Icon set
• 🔔 React Hot Toast – Toast notifications
• 🔍 react-qr-reader / html5-qrcode – QR scanner support
• 🎮 three.js – 3D graphics and scene rendering

Backend

• 🟢 Node.js – Runtime
• 🚂 Express.js – REST API framework

Database

• 🍃 MongoDB – NoSQL database (via Mongoose ODM)

AI Engine

• 🐍 Python – Core language
• 👁️ OpenCV – Computer vision
• 🎯 YOLOv8 – Object detection model

Auth

• 🔐 JWT – JSON Web Tokens for supervisor login

---

✨ Features

1. 🪖 PPE Detection System

Real-time detection of personal protective equipment using YOLOv8:

PPE Item	Detection
Helmet	✅
Safety Boots	✅
Safety Vest	✅
Safety Goggles	✅

• Input: Webcam / IP camera stream
• Output: Bounding boxes + violation alerts sent to dashboard

---

2. 📷 Multi-Camera Dashboard

For MVP, 4 laptops act as 4 camera feeds.

┌──────────┬──────────┐
│  CAM 1   │  CAM 2   │
├──────────┼──────────┤
│  CAM 3   │  CAM 4   │
└──────────┴──────────┘

Each feed runs PPE detection independently and streams results to the central dashboard.

---

3. 🪪 Worker Entry / Exit System

Each worker is assigned:

• Unique ID
• Name and Role
• QR Code for scanning

Entry Flow:

1. Worker scans QR code
2. Camera captures photo
3. System saves entry timestamp

Exit Flow:

1. Worker scans QR code
2. System saves exit timestamp

---

4. 🔒 Restricted Zone Control

Restricted zones require supervisor authorization:

Worker scans QR
       ↓
Camera checks PPE compliance
       ↓
If PPE valid → Alert sent to supervisor
"Akhil has entered Welding Zone. Approve access?"
       ↓
Supervisor: [Approve] or [Reject]

---

5. 📊 Dashboard Tables

Workers on Site

Worker	Role	Entry Time	Photo

Restricted Zone Workers

Worker	Zone	Entry Time	Status

Supervisors can assign worker roles and zone permissions from the dashboard.

---

🚀 Getting Started

Prerequisites

• Node.js >= 18
• MongoDB (local or Atlas)
• Python >= 3.8
• pip

1. Clone the Repository

git clone <your-repo-url>
cd SafeSight

2. Setup Environment Variables

cp .env.example .env
# Edit .env with your values

3. Start the Backend

cd server
npm install
npm run dev

4. Start the Frontend

cd client
npm install
npm run dev

5. Run the AI Engine

cd ai-engine
pip install -r requirements.txt
python detect_ppe_webcam.py

---

🔑 Environment Variables

MONGO_URI=mongodb://localhost:27017/safesight
JWT_SECRET=your_super_secret_jwt_key
PORT=5000

---

🗂️ Project Structure

SafeSight/
├── client/               # React frontend
│   └── src/
│       ├── components/   # Reusable UI components
│       ├── pages/        # Route pages
│       └── services/     # API service layer
├── server/               # Express backend
│   ├── config/           # DB connection
│   ├── models/           # MongoDB schemas
│   ├── routes/           # API routes
│   └── controllers/      # Business logic
├── ai-engine/            # Python PPE detection
└── docs/                 # Architecture & planning docs

---

📅 MVP Plan (3–4 Hours)

Hour	Tasks
Hour 1	Create repo, project structure, push to GitHub, setup MongoDB
Hour 2	Run PPE detection on webcam
Hour 3	Create dashboard with 4 camera feeds
Hour 4	Implement worker entry system with QR scanning

---

📋 Team Task Distribution

🖥️ Frontend Developer

• Build dashboard layout
• Implement 4-camera grid
• Build worker attendance tables
• Implement QR scanner UI
• Build restricted zone table

⚙️ Backend Developer

• Create REST APIs (workers, attendance, zones)
• Connect MongoDB with Mongoose
• Implement JWT authentication
• Create supervisor notification endpoint
• Build WebSocket for real-time alerts

🤖 AI Engineer

• Run YOLOv8 PPE detection on webcam
• Integrate webcam stream processing
• Build Flask/FastAPI endpoint for violation alerts
• Test detection accuracy

🎤 Pitch Lead

• Prepare demo video
• Train improved PPE detection model
• Prepare architecture diagrams
• Create presentation slides

---

🔗 Deployment Links

Service	URL
Frontend	(To be added)
Backend	(To be added)
Dataset	(To be added)

---

📄 License

MIT License – Built for WIET Hackverse 2.0

---

Built with ❤️ by Team Stranger Strings