Enhanced IoT-Based Face Mask Detection Framework Using Optimized Deep Learning Models A Hybrid Approach with Adaptive Algorithms

The COVID-19 pandemic has underscored the critical need for efficient, automated public health monitoring systems to ensure compliance with safety measures such as face mask usage. This project presents an Enhanced IoT-Based Face Mask Detection Framework Using Optimized Deep Learning Models: A Hybrid Approach with Adaptive Algorithms. The proposed system integrates an intelligent web-based platform with deep learning models—MobileNetV2, ResNet50, and MobileNetV3—to accurately classify individuals as "With Mask" or "Without Mask" in real time. User authentication is securely managed through a MySQL database with registration and login functionalities, ensuring controlled access to the platform. The system employs TensorFlow/Keras for model inference, image preprocessing techniques for noise reduction, and Flask as the web server to facilitate seamless image upload and prediction. Captured images are processed, classified, and displayed with corresponding labels, while the framework supports IoT integration for deployment in smart surveillance environments. The hybrid design ensures scalability, adaptability, and robust prediction performance with minimal computational resources. This framework can also be extended to other public safety monitoring applications by integrating optimized deep learning algorithms and adaptive IoT devices.

Keywords: Face Mask Detection, Deep Learning, MobileNetV2, ResNet50, MobileNetV3, Flask, TensorFlow, IoT, Image Classification, Public Safety, COVID-19 Monitoring, Real-time Prediction.