A comprehensive face mask detection system built with deep learning that can classify images of faces to determine mask-wearing status. The system uses transfer learning with MobileNetV3 architecture to efficiently identify whether individuals are wearing masks properly, improperly, or not at all.

This project implements a face mask detection system using advanced deep learning techniques, particularly relevant for public health monitoring and mask compliance verification. The system provides real-time detection capabilities through an intuitive GUI interface built with Tkinter.

• Binary Classification: Accurately detects "with_mask" vs "without_mask" status
• High Performance: Achieves 100% accuracy on test dataset
• Real-time Detection: Interactive GUI for immediate image analysis
• Transfer Learning: Leverages pre-trained MobileNetV3 for efficient training
• User-friendly Interface: Simple Tkinter-based GUI with visual feedback
• Robust Preprocessing: Comprehensive data augmentation and preprocessing pipeline

• TensorFlow/Keras: Deep learning framework for model development and training
• Kaggle Environment: Cloud platform for dataset access and model training
• Tkinter: Python GUI framework for desktop application interface
• OpenCV: Image processing and computer vision operations
• NumPy: Numerical computations and array operations
• Matplotlib: Data visualization and result plotting

The system utilizes MobileNetV3 as the backbone network, chosen for its optimal balance between accuracy and computational efficiency.

1. Input Layer

   • Accepts RGB images (224×224×3)
   • Standardized input format for consistent processing

2. Initial Convolutional Layer

   • Standard convolution (not depthwise separable)
   • 3×3 kernel with stride 2
   • Batch normalization and ReLU6 activation

3. Bottleneck Blocks

   • Core building blocks with three operations:
     • 1×1 expansion convolution
     • 3×3 depthwise convolution (spatial filtering)
     • 1×1 projection convolution
   • Residual connections for gradient flow
   • ReLU6 activation functions

4. Global Average Pooling

   • Reduces spatial dimensions to 1×1
   • Minimizes overfitting risk
   • Efficient parameter reduction

5. Fully Connected Layer

   • Dense layer for feature mapping
   • Connects to binary classification output

6. Softmax Layer

   • Probability distribution output
   • Binary classification: with_mask / without_mask

Input Image → Preprocessing → Dataset Partitioning → Training/Validation/Testing
                                       ↓
Classification ← Trained Model ← MobileNetV3 Training

1. Data Preprocessing: Image normalization, resizing, and augmentation
2. Dataset Partitioning: Train/validation/test split for robust evaluation
3. Model Training: Transfer learning with MobileNetV3 base
4. Validation: Performance monitoring during training
5. Testing: Final model evaluation on unseen data
6. Classification: Real-time prediction on new images

• Overall Accuracy: 100%
• Precision:
  • With Mask: 99%
  • Without Mask: 100%
• Recall:
  • With Mask: 100%
  • Without Mask: 99%
• F1-Score: 100% for both classes

• True Positives (with_mask): 572
• True Negatives (without_mask): 543
• False Positives: 2
• False Negatives: 3
• Total Test Samples: 1,120

The system features an intuitive Tkinter-based desktop application with:

• Image Selection: Browse and load images for analysis
• Real-time Detection: Instant mask detection with confidence scores
• Visual Feedback:
  • Red bounding box for "without_mask" (100% confidence)
  • Green bounding box for "with_mask" (100% confidence)
• Detection Statistics:
  • Total faces detected
  • Percentage breakdown of mask/no-mask detection
  • Processing status indicators

• Model loading section with file path display
• Image display area with detection overlays
• Results panel showing detection statistics
• Control buttons for image selection and detection execution

1. Clone the Repository

   git clone https://github.com/esraamhmd/FaceMaskDetector.git
   cd FaceMaskDetector

2. Install Dependencies

   pip install tensorflow opencv-python tkinter numpy matplotlib pillow

3. Download Pre-trained Model

   • Ensure the trained model file (face_mask_mobilenetV3.h5) is in the project directory
   • Model can be downloaded from the releases section or trained using the provided notebook

python face_mask_detector_gui.py

1. Load Model: Click "Load Model" to initialize the pre-trained model
2. Select Image: Choose an image file for mask detection
3. Detect Masks: Click "Detect Masks" to analyze the selected image
4. View Results: Check detection results in the statistics panel

• JPEG (.jpg, .jpeg)
• PNG (.png)
• BMP (.bmp)
• TIFF (.tiff)

FaceMaskDetector/
├── face_mask_detector_gui.py    # Main GUI application
├── model_training.ipynb         # Training notebook
├── face_mask_mobilenetV3.h5     # Trained model weights
├── requirements.txt             # Project dependencies
├── README.md                   # Project documentation
└── sample_images/              # Test images directory

The model training process includes:

1. Data Augmentation: Rotation, scaling, brightness adjustment
2. Transfer Learning: Fine-tuning MobileNetV3 pre-trained weights
3. Optimization: Adam optimizer with learning rate scheduling
4. Regularization: Dropout and batch normalization for generalization
5. Early Stopping: Prevents overfitting during training

• Public Health Monitoring: Automated mask compliance checking
• Security Systems: Integration with surveillance cameras
• Educational Tools: Demonstration of computer vision concepts
• Research Platform: Base for advanced face detection research

The system can be extended for:

• Multi-class Detection: Proper/improper mask wearing classification
• Real-time Video: Webcam integration for live detection
• Mobile Deployment: Conversion to mobile-friendly formats
• API Integration: RESTful service for web applications

1. Fork the repository
2. Create a feature branch (git checkout -b feature/new-feature)
3. Commit your changes (git commit -am 'Add new feature')
4. Push to the branch (git push origin feature/new-feature)
5. Create a Pull Request

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

Esraa Mohammed - GitHub Profile

• TensorFlow team for the MobileNetV3 architecture
• Kaggle community for dataset resources and compute environment
• OpenCV contributors for computer vision tools
• Public health organizations inspiring this COVID-19 era project

For questions, suggestions, or collaborations:

• Open an issue on GitHub
• Contact the repository owner through GitHub

⭐ Star this repository if you find it helpful for your projects!