Classroom Attendance System

This project implements an automated classroom attendance system using face recognition.

Setup

1. Clone the repository
2. Create a virtual environment: python -m venv venv
3. Activate the virtual environment
4. Install dependencies: pip install -r requirements.txt
5. Download required dlib models (see README for details)
6. Run the script: python classroom_attendance.py

Requirements

• Python 3.7+
• See requirements.txt for Python package dependencies
• dlib face recognition models (not included in repo due to size)

Usage

[Add usage instructions here]

License

[Add license information here]

Dlib Models

This project requires two dlib models which are not included in the repository due to their size:

1. dlib_face_recognition_resnet_model_v1.dat
2. shape_predictor_68_face_landmarks.dat

To download these models:

1. Visit the dlib model download page: http://dlib.net/files/
2. Download the following files:
   • shape_predictor_68_face_landmarks.dat.bz2
   • dlib_face_recognition_resnet_model_v1.dat.bz2
3. Extract the .bz2 files to get the .dat files
4. Place both .dat files in the project root directory

Note: These models are essential for face recognition and landmark detection. The project will not run without them.

Using Webcam

By default, the script is set to use your computer's webcam. If you want to use a different camera or switch back to an RTSP stream:

1. Open classroom_attendance.py
2. Locate the line cap = cv2.VideoCapture(0)
3. Change 0 to the index of your desired camera (e.g., 1 for a second webcam)
4. For RTSP stream, replace the line with:

   cap = cv2.VideoCapture("rtsp://your_rtsp_stream_url_here")
   

Preparing Known Faces

To use face recognition with multiple images per person:

1. Create a directory named known_faces_images in the project root.
2. Add clear, front-facing images of known individuals to this directory. Name each image file as follows: "person_name_uniqueidentifier.jpg" (e.g., "john_doe_1.jpg", "john_doe_2.jpg", "jane_smith_1.jpg")
3. Run the face encoding generator:python generate_known_faces.py
4. This will create known_faces.pkl, which contains average encodings for each person.

Note:

• Ensure you have permission to use individuals' images for this purpose.
• Using multiple images per person can improve recognition accuracy.
• The script will calculate an average encoding for each person from their multiple images.

Environment Setup

This project is developed using Python 3.12.2. To set up the environment:

1. Ensure you have Python 3.12.2 installed
2. Create a virtual environment:python -m venv venv
3. Activate the virtual environment:

• Windows: venv\Scripts\activate
• MacOS/Linux: source venv/bin/activate

4. Install the required packages:pip install -r requirements.txt Note: If you encounter issues with dlib installation, refer to the Troubleshooting section below.

Troubleshooting

If you encounter issues installing dlib:

1. Ensure you have CMake installed: pip install cmake,
2. Install Visual Studio Build Tools 2019 or later with "Desktop development with C++" workload, then 2a. Try installing dlib:
3. If issues persist, download a pre-built wheel from Dlib-Wheels and install it manually:
4. For RetinaFace, you might need to install it from the GitHub repository (pip install git+https://github.com/serengil/retinaface.git)

Recent Updates

• Fixed issues with OpenCV tracking methods
• Improved face detection using RetinaFace
• Simplified the main script for better performance
• Added FPS counter to the GUI

Recent Updates

• Added comprehensive face analysis (emotion, age, gender)
• Implemented attentiveness detection based on emotional state
• Improved display stability for face information
• Optimized face detection and analysis frequency
• Enhanced multi-threading for better performance

Usage

1. Ensure your webcam is connected and functioning.
2. Run the script: python classroom_attendance.py
3. Click the 'Start' button to begin face detection and analysis.
4. The system will display:
   • Bounding boxes around detected faces
   • Name and confidence of recognized individuals
   • Emotion, age, and gender for each detected face
   • Attentiveness status based on emotional state
5. Click 'Stop' to pause the detection and 'Exit' to close the application.

Note: Face analysis is performed periodically to balance performance and accuracy.

Recent Updates (MediaPipe Integration)

• Replaced YOLO face detection with MediaPipe Face Detection for improved accuracy and performance
• Streamlined the face detection process
• Removed the maximum face limit for more flexible detection
• Adjusted minimum face size for better detection of smaller faces
• Optimized the main loop for smoother performance

MediaPipe Face Detection

The latest version of the script (Media_pipe.py) uses MediaPipe Face Detection instead of YOLO. This change offers several benefits:

• Improved accuracy in face detection
• Faster processing, leading to better real-time performance
• More stable detection across various lighting conditions and face orientations

Usage of Media_pipe.py

1. Ensure you have installed the additional dependency: pip install mediapipe
2. Run the script: python Media_pipe.py
3. The GUI will open, showing the video feed from your webcam.
4. Use the 'Start' button to begin face detection and recognition.
5. Adjust the 'Confidence Threshold' slider to fine-tune detection sensitivity.
6. Select a specific recognition model or use "All Models" for ensemble recognition.
7. The system will display:

• Bounding boxes around detected faces
• Names and confidence scores of recognized individuals
• Real-time face count and FPS (Frames Per Second)

8. Use the 'Stop' button to pause detection and 'Exit' to close the application.

Configuration

• MIN_FACE_SIZE: Minimum size (in pixels) for a face to be considered for detection (default: 10)
• CONFIDENCE_THRESHOLD: Minimum confidence score for face detection (adjustable in GUI, default: 0.5)
• ANALYSIS_FREQUENCY: How often to perform full face analysis (adjustable in GUI)

Performance Notes

• The MediaPipe face detection model is optimized for real-time performance on CPU.
• Face recognition is still performed using DeepFace models, which may be computationally intensive.
• Adjust the analysis frequency and confidence threshold as needed to balance accuracy and performance.

Known Issues

• Face recognition accuracy may vary depending on lighting conditions and face angles.
• High CPU usage may occur during face recognition processes.

Recent Updates (Web-based GUI)

• Migrated from PySimpleGUI to a web-based interface using Flask
• Implemented a responsive layout with video feed and settings side-by-side
• Added real-time controls for face recognition parameters
• Improved user experience with a more intuitive interface

Setup

1. Clone the repository
2. Create a virtual environment: python -m venv venv
3. Activate the virtual environment
4. Install dependencies: pip install -r requirements.txt
5. Download required dlib models (see Dlib Models section for details)
6. Run the Flask application: python app.py

Requirements

• Python 3.7+
• See requirements.txt for Python package dependencies
• dlib face recognition models (not included in repo due to size)
• Flask for web server

Usage

1. Start the Flask server by running python app.py
2. Open a web browser and navigate to http://localhost:5000
3. Use the web interface to:
   • Start/stop face recognition
   • Adjust confidence threshold and minimum face size
   • Select recognition models
   • Toggle mood and attentiveness display
   • Save attendance data
   • Reload known faces
   • Train custom ensemble model

Web Interface Features

• Live video feed display
• Real-time face detection and recognition
• Adjustable settings for face detection parameters
• Buttons for various system actions (start, stop, save attendance, etc.)
• Responsive layout for different screen sizes