📸 Real-Time Face Recognition Attendance System

A real-time attendance tracking system using facial recognition, powered by FaceNet embeddings, KNN classifier, and OpenCV. The system identifies individuals through a live webcam feed and marks their entry and exit times with high accuracy. Unknown individuals are automatically detected and their images are saved with timestamps.

---

🚀 Features

• 🔍 Real-time face detection using webcam.
• 🤖 Face recognition using FaceNet + KNN classifier.
• ✅ Automatic attendance marking with entry and exit timestamps.
• 🔐 Confidence threshold to prevent false positives.
• 🧠 Face stabilization logic to avoid flickers.
• 🧾 Attendance logging in a structured CSV file.
• ❌ Unknown face detection with red bounding box and snapshot saving.
• 🖥️ GUI-based control to start/stop camera via Tkinter.

---

🛠️ Tech Stack

• Python 3.x
• OpenCV
• Keras-FaceNet
• NumPy
• Pandas
• Tkinter
• KNN Classifier (trained model loaded via pickle)

---

📌 How to Use

1. Prepare Dataset

• Organize person images in folders labeled with person IDs (e.g., 001, 002, etc.).
• Each folder should contain 6–7 facial images with varied expressions and angles.

2. Train Model

• Train and save the KNN model (knn_model.pkl) and label encoder (label_encoder.pkl) using FaceNet embeddings.

3. Update Attendance File

• Ensure Attendance.csv has the following columns:

4. Control Camera

• Run the script.
• A simple GUI opens with a "Stop Camera" button to terminate the session.

5. View Reports

• After stopping the camera, a file named Attendance_Report.csv is saved with:

6. Unknown Detection

• Unknown individuals:
• Are displayed with a red bounding box.
• Are labeled as “Unknown”.
• Have their snapshot saved to unknown_snapshots/ with a timestamped filename.

---

📷 Sample Output

✅ Known Face Detection:

• Green bounding box
• Name label displayed
• Attendance marked automatically

❌ Unknown Face Detection:

• Red bounding box
• “Unknown” label
• Snapshot saved to unknown_snapshots/unknown_YYYYMMDD_HHMMSS.jpg

---

🧠 Future Enhancements

• Dynamic KNN training from the UI.
• Face re-registration module.
• Integration with databases (MySQL, Firebase).
• Email alerts for unknown detections.
• Multi-camera support.

---

📄 License

This project is open-source and free to use for educational and research purposes.

---

🤝 Acknowledgements

• FaceNet by Keras-Facenet
• OpenCV for real-time video capture
• Inspiration from practical AI applications in smart surveillance and biometric systems

---

👤 Contributors

• Dogga Pavan Sekhar
• Koyyada Shreeshanth

---

Feel free to contact shreeshanthgoud@gmail.com, doggapavansekhar@gmail.com