🚨 SafeSphere AI

Predictive Crowd Intelligence System for Large-Scale Events

SafeSphere AI is an AI-powered system designed to predict, prevent, and manage crowd risks in large-scale events like the Kumbh Mela. It enables real-time decision-making to avoid stampedes, optimize crowd movement, and improve public safety.

---

🌍 Problem Statement

Large gatherings often suffer from:

• Sudden crowd surges
• Poor real-time visibility
• Lack of predictive intelligence
• High risk of stampedes

Traditional systems are reactive, responding only after congestion occurs.

---

💡 Solution

SafeSphere AI introduces a predictive AI command system that:

• Monitors crowd density across zones
• Detects high-risk and critical areas
• Predicts crowd buildup trends
• Suggests dynamic rerouting strategies
• Generates real-time alerts

---

🧠 Key Features

• 📊 Crowd Density Analysis
• ⚠️ Risk Detection (High / Critical / Stampede Risk)
• 🔮 Predictive Intelligence (Trend Forecasting)
• 🔀 Smart Crowd Rerouting
• 📢 Real-Time Alerts
• 🧩 Structured JSON API for easy integration

---

🏗️ System Architecture

Simulated Data → AI Agent (Antigravity + Gemini) → Cloud Run API → JSON Response

🔗 Google Cloud Integration

• Deployed on Google Cloud Run (serverless, scalable)

• Uses Gemini API for AI decision-making

• Designed for integration with:

  • Google Firestore (real-time data storage)
  • Pub/Sub (event-driven updates)
  • Cloud Logging (monitoring & debugging)

---

⚙️ Efficiency & Performance

• Lightweight processing for low-latency responses
• Optimized decision logic to minimize compute overhead
• Designed for real-time scalability using serverless architecture
• Supports future enhancements like caching and batch processing

---

🔐 Security

• API keys managed via environment variables
• Secure deployment using Google Cloud infrastructure
• Designed with controlled access patterns

---

♿ Accessibility

• Simple and intuitive REST API
• Structured JSON responses for easy consumption
• Clear error handling and validation
• Designed for integration with dashboards and assistive systems

---

📡 API Usage

Endpoint

POST /analyze

Request Example

{
  "zones": [
    {"zone_id": "A1", "density": 85, "movement_speed": 1.2},
    {"zone_id": "B2", "density": 95, "movement_speed": 0.4}
  ]
}

Response Example

{
  "zone_status": [
    {
      "zone_id": "B2",
      "density_level": "Critical",
      "risk_level": "Stampede Risk",
      "trend": "Increasing"
    }
  ],
  "actions": [
    "Redirect crowd from Zone B2 to Zone C1"
  ],
  "alerts": [
    "Avoid Zone B2 due to heavy congestion"
  ]
}

---

🧪 Testing

Basic test coverage is implemented to validate core functionality.

Example Test Case

def test_analyze_endpoint():
    response = client.post("/analyze", json={
        "zones": [{"zone_id": "A1", "density": 90, "movement_speed": 0.5}]
    })
    assert response.status_code == 200

Test Coverage Includes:

• ✅ Normal scenarios
• ⚠️ High-risk scenarios
• ❗ Edge cases (empty input / invalid data)

---

🛠️ Setup & Deployment

1. Clone Repository

git clone https://github.com/YOUR_USERNAME/safesphere.git
cd safesphere

2. Install Dependencies

pip install -r requirements.txt

3. Configure Environment

GEMINI_API_KEY=your_api_key_here

4. Run Locally

python app.py

5. Deploy to Cloud Run

gcloud run deploy safesphere-api --source .

---

🎯 Use Cases

• Religious gatherings (e.g., Kumbh Mela)
• Sports stadiums
• Concerts and festivals
• Smart city crowd management

---

📈 Future Enhancements

• Real-time CCTV integration
• IoT sensor-based tracking
• Live dashboard with heatmaps
• Automated emergency response system
• Multi-agent orchestration

---

📌 Disclaimer

This is a prototype system using simulated data, designed for demonstration and scalability.

---

👨‍💻 Author

Developed by Shiva Tiwari

---

⭐ Support

If you find this project useful, consider giving it a ⭐ on GitHub!