Face detection based on K-medoids clustering and associated with convolutional neural networks.

Area of Science:

• Computer Science
• Artificial Intelligence
• Public Health

Background:

• The COVID-19 pandemic necessitated new public health measures, including mandatory face mask usage.
• Manual monitoring of face mask compliance in public spaces is inefficient and challenging.
• Automated solutions are crucial for effective face mask identification and public conduct monitoring.

Purpose of the Study:

• To develop and evaluate novel methods for automated face mask detection.
• To improve the accuracy and efficiency of identifying individuals not adhering to face mask policies.
• To enhance public safety and reduce the spread of infectious diseases like COVID-19.

Main Methods:

• Image pre-processing techniques including K-medoids, K-means, and Fuzzy K-Means (FKM) were employed to enhance facial image quality and reduce noise.
• Machine learning models, specifically Convolutional Neural Networks (CNNs) with pre-trained DenseNet201, VGG-16, and VGG-19, alongside Support Vector Machine (SVM), were investigated for face mask detection.

Main Results:

• The K-medoids clustering algorithm combined with the DenseNet201 pre-trained model achieved a high accuracy of 97.7% for face mask identification.
• Image segmentation techniques were found to significantly improve the accuracy of face mask detection.
• The developed tool demonstrated effectiveness in identifying face masks even from a side-on perspective.

Conclusions:

• Automated face mask detection systems are a viable and effective solution for monitoring public compliance.
• The proposed K-medoids and DenseNet201 approach offers a robust method for accurate face mask identification.
• Further development in image analysis and machine learning can enhance the utility of such tools in public health surveillance.