😷 Face Mask Detection & Tracking System

📌 Project Overview

This project is an end-to-end Computer Vision system designed to automate the monitoring of face mask compliance in real-time. Unlike standard detection models, this system integrates Object Tracking logic to assign unique IDs to individuals. This ensures accurate counting for statistical analysis and prevents duplicate entries in the dashboard.

The solution is lightweight, efficient, and deployed as an interactive web application using Streamlit.

🚀 Key Features

• Real-time Detection: Instantly detects faces and classifies them as "With Mask" or "Without Mask".
• Smart Tracking System: Implements tracking algorithms to maintain unique IDs for subjects across video frames, ensuring accurate "People Counting".
• Analytics Dashboard: A dynamic Streamlit interface displaying live stats and counters.
• High Efficiency: Uses MobileNetV2 for fast inference, making it suitable for edge devices and surveillance systems.

📊 Dataset

The model was trained on a balanced dataset of approximately 12,000 images:

• ~6,000 With Mask: Scraped from Google Search.
• ~6,000 Without Mask: Preprocessed from the CelebFace dataset.
• Preprocessing: Images were resized to 224x224, normalized, and augmented (Rotation, Zoom, Horizontal Flip) to prevent overfitting.

🧠 Model Performance

We experimented with three Transfer Learning architectures. MobileNetV2 was selected as the final model due to its superior balance between accuracy and speed (FPS).

Model Architecture	Validation Accuracy	Validation Loss	Status
VGG16	99.50%	0.0179	High Accuracy / Slow Inference
ResNet50	77.38%	0.5294	Underperformed
MobileNetV2	99.25%	0.0191	Selected (Fastest & Accurate)

Result: The MobileNetV2 model achieved 99.25% accuracy with minimal loss, making it ideal for real-time video processing.

💻 Tech Stack

• Deep Learning: TensorFlow, Keras.
• Computer Vision: OpenCV.
• Web Framework: Streamlit.
• Language: Python.

📸 Demo

🛠️ Installation & Usage

1. Clone the repository:

   git clone https://github.com/omarDlgaber/Face_Mask_Detection.git
   cd Face_Mask_Detection

2. Install dependencies:

   pip install -r requirements.txt

3. Run the application:

   streamlit run app.py

👤 Author

Omar Adel

• Project: Final Project | Data Science & AI Diploma
• Date: December 2025

---

⭐ If you find this project useful, please consider giving it a star on GitHub!