🛡️ Intelligent Smart Surveillance System

An AI-powered real-time surveillance system that integrates:

• 🧍 Human detection
• 🤸 Pose-based behavior analysis
• 🔪 Hazardous object (weapon) detection
• 🏃 Speed and motion estimation

All combined into a live monitoring dashboard for intelligent surveillance.

This project demonstrates building a real-time AI system with backend integration and automated decision-making, aligned with production AI workflows.

---

🚀 Features

• 🎥 Real-time video streaming (Webcam)

• 🧍 Human detection using YOLOv8

• 🤸 Pose estimation using MediaPipe

• 🚨 Fall detection & abnormal behavior detection

• 🏃 Running / high-speed movement detection

• 🔪 Hazardous object detection (knife, tools, etc.)

• 📏 Speed estimation using motion tracking

• 📊 Live dashboard with alerts and system stats

• 🔔 Alert logging system

• 📁 Data storage:

  • Pose sequences
  • Alerts
  • Reports

---

📸 Demo Preview

🧠 Live AI Detection (Normal Activity)

---

🏃 Behavior Detection (Running)

---

🔪 Weapon Detection Alert

---

📊 System Statistics Dashboard

---

⚠️ Weapon Detection Logs

---

🤸 Behavior Analysis Panel

---

🚨 Security Alerts Panel

---

🧠 Tech Stack

• Python 3.10 (recommended via Anaconda)
• Flask (Backend + API)
• OpenCV (Video processing)
• YOLOv8 (Ultralytics)
• MediaPipe (Pose estimation)
• NumPy, Pandas, Scikit-learn
• HTML, CSS, JavaScript (Frontend)

---

📁 Project Structure

EnhancedSmartSurveillance/
│
├── app.py
├── object_detector.py
├── speed_estimator.py
├── label_generator.py
├── knife.yaml
│
├── templates/
│   └── index.html
│
├── static/
│
├── data/
│   ├── pose_sequences/
│   ├── alerts/
│   └── reports/
│
├── models/
│
├── requirements.txt
└── README.md

---

⚙️ Setup Instructions

1️⃣ Clone the Repository

git clone https://github.com/payaldongre/EnhancedSmartSurveillance.git
cd EnhancedSmartSurveillance

---

2️⃣ Create Environment

conda create --prefix ./env python=3.10
conda activate ./env

---

3️⃣ Install Dependencies

pip install -r requirements.txt

---

4️⃣ Fix PyTorch Compatibility (IMPORTANT)

pip install torch==2.5.1 torchvision==0.20.1 torchaudio==2.5.1 --extra-index-url https://download.pytorch.org/whl/cpu

---

▶️ Run the Project

# Navigate to project folder
cd EnhancedSmartSurveillance

# Activate environment
conda activate ./env

# Run application
python app.py

---

🌐 Access Dashboard

After running, open:

• Local:

http://localhost:5000

• Same network (optional):

http://<your-ip-address>:5000

Example:

http://192.168.1.5:5000

---

📡 API Endpoints

Endpoint	Description
/	Dashboard
/video_feed	Live video stream
/api/stats	System statistics
/api/alerts	Alert logs
/api/weapon_detections	Weapon alerts
/api/behavior_history	Behavior tracking
/api/test_alert	Trigger test alert
/api/clear_alerts	Reset alerts

---

🎯 Detection Capabilities

🧍 Human Detection

• YOLOv8-based detection

🤸 Behavior Analysis

• Normal activity
• Abnormal behavior
• Running detection
• Fall detection

🔪 Hazard Detection

• Detects:

  • Knife
  • Tools / objects

• Triggers alerts when near a person

🏃 Speed Estimation

• Motion tracking across frames

• Based on:

  • Pixel displacement
  • Pose normalization
  • Temporal smoothing

---

⚠️ Important Notes

• Python 3.10 recommended (avoid 3.13)
• Webcam must be available
• Close other apps using camera

👉 YOLO weights will be downloaded automatically on first run.

---

🧪 Testing

• Start the system and open dashboard

• Try:

  • Fast movement → running detection
  • Sudden fall → fall detection
  • Object in hand → weapon detection

---

📌 Future Scope

• ESP32-CAM integration
• Multi-camera support
• Cloud deployment
• SMS / Email alerts
• Database integration

---

👩‍💻 Contributors

• Payal Dongre
• Priyanka Jadhav

---

🏁 Summary

A real-time intelligent surveillance system combining:

• Object detection
• Pose estimation
• Behavior analysis
• Motion tracking

into a unified monitoring solution.

---

Note: Large files, model weights, and environment folders are excluded using .gitignore for efficient repository management.