CrowdSense-AI

Real-time crowd detection using machine learning for people detection and counting from video streams. This project identifies human presence in frames and visualizes detections in real time.

Built using Google AI Studio and TensorFlow.js.

---

🚀 Features

• Real-time people detection
• Crowd / people counting
• Bounding box visualization
• Browser-based execution
• Lightweight and fast inference
• Video stream support

---

🧠 Tech Stack

• TensorFlow.js
• TypeScript
• Vite
• HTML/CSS
• Node.js

---

📂 Project Structure

CrowdSense-AI/
│
├── src/              # Core detection logic
├── scripts/          # Utility scripts
├── docs/             # Documentation files
├── index.html        # App entry point
├── package.json      # Dependencies
├── vite.config.ts    # Vite configuration
└── README.md

---

⚙️ How It Works

1. Video input is captured from camera or file
2. TensorFlow.js loads the detection model
3. Humans are detected in each frame
4. Bounding boxes are drawn around people
5. Total count is calculated
6. Results displayed in real time

---

🖥️ Run Locally

Prerequisites

• Node.js installed

Installation

npm install

Set API Key

Create .env.local and add:

GEMINI_API_KEY=your_api_key_here

Run Development Server

npm run dev

App will start at:

http://localhost:5173

---

📊 Use Cases

• Smart surveillance systems
• Event monitoring
• Campus safety
• Public space analytics
• Footfall counting

---

🔮 Future Improvements

• Heatmap visualization
• Person tracking IDs
• Multi-camera support
• Overcrowding alerts

---

👤 Author

Apurva Nikam