🔒 Smart Surveillance with YOLOv8 + ArcFace + SQLite

This project implements a real-time smart CCTV system that combines YOLOv8 object detection and ArcFace embeddings for robust face recognition, with SQLite database integration for identity management and event logging.

---

🚀 Features

• YOLOv8 for detecting people in video streams (webcam, IP camera, or pre-recorded video).
• Face extraction from detected persons.
• ArcFace embeddings (via DeepFace) to compute unique 512-D face vectors.
• SQLite database for:
  • Enrolling new users.
  • Storing embeddings.
  • Logging recognition events with timestamps + confidence scores.
• Outputs:
  • Annotated live video with bounding boxes + labels.
  • SQLite log records.
  • Future hooks: Telegram/Email alerts.

---

📐 System Architecture

---

📊 Sample Output

1. Real-time Recognition

• The system detects multiple people and matches them with enrolled users.
• Example: Alice (0.91) and Bob (0.88) detected in a live stream.

2. Event Logs

SQLite table logs every recognition event:

• id, name, confidence, timestamp.

---

📁 Repository Structure

minimal-yolov8-arcface-repo/
 ├── app.py                     # YOLOv8 + ArcFace real-time script
 ├── minimal_yolov8_arcface.ipynb # Notebook demo with SQLite
 ├── requirements.txt           # Dependencies
 ├── README.md                  # Documentation
 ├── LICENSE                    # MIT License
 ├── assets/                    # Screenshots + diagrams
 │    ├── architecture_pro.png
 │    ├── sample_output.png
 │    └── logs_preview.png
 └── enroll/                    # Enrolled faces
      ├── Alice/
      └── Bob/

---

⚡ Quick Start

1. Install dependencies:

   pip install -r requirements.txt

2. Enroll identities (place face images inside enroll/<Name>/).
3. Run the real-time script:

   python app.py

4. Or explore the notebook:

   jupyter notebook minimal_yolov8_arcface.ipynb

---

📜 License

This project is licensed under the MIT License.