Sentinel Facial Recognition Attendance System

Overview

The Sentinel Facial Attendance System is a real-time facial recognition-based attendance management system. It eliminates the need for manual attendance tracking by leveraging computer vision and machine learning techniques to automatically mark attendance. The system is designed for schools, colleges, and corporate offices to enhance accuracy and efficiency.

This system uses OpenCV's pre-trained Haar Cascade Classifier to detect faces from a live camera feed and compare them against stored images to identify individuals. It records attendance and captures images of unauthorized persons separately. The system allows adding new students and retraining the model dynamically.

---

Key Features

• Real-Time Facial Recognition: Identifies individuals from live video streams.
• Attendance Marking: Logs attendance efficiently and accurately.
• Intruder Detection: Captures and stores images of unknown persons and sends email alerts.
• Automated Data Logging: Stores attendance data in CSV format for easy analysis.
• Training and Retraining: Supports adding new individuals and updating the model.
• Easy Setup: Minimal configuration required.
• Customizability: Modify training images and detection settings as needed.

---

Technology Stack

• Programming Language: Python 3
• Computer Vision Library: OpenCV
• Machine Learning Algorithm: Haar Cascade Classifier
• Data Storage: CSV files for attendance logs
• Email Alerts: SMTP for sending intruder detection alerts
• Environment Management: Virtual Environment (venv)

---

Understanding Haar Cascade Classifier

The Haar Cascade Classifier is a machine learning object detection algorithm used to detect objects in images or video. It works by analyzing patterns in an image and comparing them to pre-trained patterns stored in an XML file.

How It Works

1. Feature Extraction: Haar-like features (patterns) are extracted from an image.
2. Integral Image Calculation: The algorithm quickly calculates feature values.
3. Adaboost Training: Weak classifiers are combined into a strong classifier.
4. Cascade Structure: The classifier works in multiple stages, rejecting non-face regions early to optimize speed.

---

Comparison with Alternative Methods

Method	Accuracy	Speed	Complexity	Requires Training?
Haar Cascade (Used in Sentinel)	High	Fast	Low	No (Pre-trained Model)
HOG + SVM	Medium	Medium	Medium	Yes
Deep Learning (CNNs, YOLO, etc.)	Very High	Slow	High	Yes

Why Haar Cascade?

• Fast: Uses a cascade approach for quick processing.
• Lightweight: Requires minimal computational power.
• Pre-trained: Works out of the box with minimal setup.

---

Installation Guide

1. Clone the Repository

git clone https://github.com/yourusername/Sentinel-Facial-Attendance-System.git
cd Sentinel-Facial-Attendance-System

2. Create a Virtual Environment

python -m venv venv

Activate the virtual environment:

• Windows:

  venv\Scripts\activate

• Linux/macOS:

  source venv/bin/activate

3. Install Dependencies

pip install -r requirements.txt

4. Download Pre-trained Face Detection Model

Ensure that the Haar Cascade XML file (haarcascade_frontalface_default.xml) is in the project root directory. If not, download it from OpenCV GitHub and place it in the project directory.

---

Usage Guide

1. Adding New Students

• Place training images in the TrainingImage/ folder.
• Use the following naming convention for images:

  <StudentID>_<StudentName>.jpg
  

• Update StudentDetails.csv in the StudentDetails/ folder.

2. Running the Attendance System

python main.py

3. Viewing Attendance Records

• Attendance records are stored in the Attendance/ folder as CSV files.
• Each file is named with the current date and contains the following details:
  • Student ID
  • Name
  • Date
  • Time

4. Intruder Detection and Email Alerts

When an unknown face is detected:

1. The system captures the image and saves it in the Intruder/ folder.
2. An email alert is sent to the administrator with the detected image attached.

Setting Up Email Alerts

• Configure the SMTP server in the config.py file:

  SMTP_SERVER = "smtp.gmail.com"
  SMTP_PORT = 587
  EMAIL_SENDER = "your_email@gmail.com"
  EMAIL_PASSWORD = "your_password"
  EMAIL_RECEIVER = "admin_email@gmail.com"

• Ensure that less secure apps access is enabled on your email account.

---

Troubleshooting

1. haarcascade_frontalface_default.xml Not Found

Ensure the file is in the project root directory. If missing, download and place it manually.

2. cv2.CascadeClassifier AttributeError

If OpenCV is not installed correctly, reinstall it:

pip uninstall opencv-python opencv-contrib-python
pip install opencv-python opencv-contrib-python

3. Camera Not Detected

Check if your webcam is properly connected and accessible.

4. No Face Detected

• Adjust camera angle and lighting conditions.
• Ensure the face is clearly visible to the camera.

5. Email Alerts Not Working

• Verify SMTP server settings.
• Check if your email account allows less secure apps.
• Ensure the internet connection is stable.

---

Contributing

We welcome contributions! Fork the repository, make your changes, and submit a pull request.

---

License

This project is licensed under the MIT License. See the LICENSE file for details.

---

Acknowledgements

• OpenCV for providing robust computer vision tools.
• Python for its simplicity and power.
• Haar Cascade Classifier for efficient face detection.

---

Author

• Dev Sharma - GitHub Profile