Abstract

This project aims to develop an automated face mask detection system using deep learning techniques. The system will identify whether a person in an image is wearing a mask or not wearing a mask. A Convolutional Neural Network (CNN) model will be trained on a dataset containing face images with and without masks. The model will be integrated with a real-time detection interface to monitor live camera feeds or uploaded images. This technology can be applied in public safety monitoring, hospital check-ins, and smart surveillance systems.

Functional Requirements

1. Data Collection & Preprocessing

1. Collect or create a dataset containing face images with and without masks.
2. Apply face detection using OpenCV or MediaPipe.
3. Resize and normalize images for model compatibility.
4. Perform data augmentation (e.g., rotation, flipping, brightness adjustment) to improve robustness.

2. Model Development

3. Implement a Convolutional Neural Network (CNN) or fine-tune a pretrained model (such as MobileNet or VGG16).
4. Train the model to classify images into two categories: Mask and No Mask.
5. Use dropout, batch normalization, and hyperparameter tuning to optimize performance.
6. Target validation accuracy above 90%.

3. User Interface Development

Desktop Application (Tkinter/PyQt):

• Allow users to upload photos for detection.
• Display prediction results and confidence scores.

Web Application (Flask/Django):

• Provide a web-based interface for image uploads or live camera feed integration.

4. Real-Time Detection

• Use OpenCV to capture live video from a webcam or CCTV feed.
• Detect and classify faces as Mask or No Mask in real-time.
• Display bounding boxes and labels with confidence scores.

5. Database Integration

• Store detection results including timestamps, image paths, and predictions using SQLite or MySQL.
• Allow administrators to view and analyze detection logs.

6. Performance Evaluation & Optimization

• Evaluate system performance using metrics such as accuracy, precision, recall, and F1-score.
• Test under various lighting and camera conditions.
• Optimize the model for faster real-time performance while maintaining high accuracy.

Tools and Technologies Required

Programming Language:

Python

Libraries & Frameworks:

• Deep Learning: TensorFlow / Keras or PyTorch
• Image Processing: OpenCV, Pillow, dlib
• Backend: Flask / Django
• GUI: Tkinter / PyQt

Dataset:

• Kaggle Face Mask Dataset (https://www.kaggle.com/datasets/omkargurav/face-mask-dataset)
• Custom images (optional for fine-tuning)

Database:

SQLite / MySQL

IDE:

PyCharm, VS Code, or Jupyter Notebook

Hardware:

• CPU: Intel i5 or AMD Ryzen 5 and above
• RAM: Minimum 4GB
• GPU (optional for faster training)
• Storage: A few GBs for dataset and model weights