Smart Surveillance using 5G Technology

🔹 Problem Statement

Traditional surveillance systems face challenges such as:

• High latency in live video streaming.
• Limited ability to process data in real time.
• Inefficient anomaly detection requiring manual monitoring.
• High bandwidth consumption when transmitting raw video continuously.
• Security concerns with unencrypted or poorly managed camera feeds.

👉 To address these challenges, we propose Smart Surveillance powered by 5G.
Our system leverages edge computing, AI/ML-based anomaly detection, and secure data storage to detect threats in real time and notify administrators immediately — transmitting only anomalous frames instead of continuous video.

---

🔹 Key Technological Aspects

1. 5G Integration

   • High bandwidth and ultra-low latency.
   • Supports real-time anomaly detection and rapid alerts.

2. Edge Computing

   • Video frames are processed locally (on the camera or nearby edge server).
   • Reduces cloud dependency and improves response time.

3. AI/ML Models

   • Uses YOLOv5 (PyTorch) for object detection.
   • Detects anomalies such as weapons, fire, suspicious objects, and fights.
   • Configurable anomaly classes (ANOMALY_CLASSES).

4. Database Integration

   • PostgreSQL is used to log anomalies.
   • Stores anomaly name, confidence score, timestamp, and frame (as binary data).
   • Can be extended to store file paths or integrate with dashboards.

5. Cybersecurity

   • Secure storage of database credentials using .env files.
   • Future scope: add encryption for data transmission (TLS/SSL).

6. Visualization & Alerts

   • Annotated video feed shows real-time detections.
   • Console alerts highlight anomalies with timestamps.
   • Frames containing anomalies are stored in PostgreSQL for auditing.

---

🔹 Project Workflow

1. 5G-enabled cameras capture live footage.
2. Edge computing (local inference with YOLOv5) processes video in real time.
3. AI/ML anomaly detection identifies threats instantly.
4. System triggers alerts and saves anomalies into PostgreSQL.
5. Cloud/DB storage preserves frames for audit and retraining.
6. Admins review anomalies without needing to sift through hours of video.

---

🔹 Tech Stack

• Programming Language: Python
• Deep Learning Framework: PyTorch
• Object Detection Model: YOLOv5 (Ultralytics)
• Database: PostgreSQL 17
• Environment Management: dotenv
• Libraries: OpenCV, psycopg2, requests

---

🔹 Installation & Setup

1️⃣ Clone Repository

git clone https://github.com/your-username/5g-smart-surveillance.git
cd 5g-smart-surveillance