An effective stacked autoencoder based depth separable convolutional neural network model for face mask detection.

Area of Science:

• Computer Vision
• Machine Learning
• Public Health

Background:

• The COVID-19 pandemic necessitated widespread adoption of public health measures, including mask-wearing.
• Effective screening systems are crucial for public spaces to ensure compliance with mask mandates.
• Existing face detection models often lack integration with dimensionality reduction techniques.

Purpose of the Study:

• To develop a deep learning model for accurate detection of mask usage and proper wearing.
• To address the need for automated screening systems in public areas.
• To improve upon existing face detection methods by incorporating dimensionality reduction.

Main Methods:

• Implementation of a Stacked Auto Encoder (SAE) technique.
• Integration of Principal Component Analysis (PCA) for feature reduction.
• Utilizing Depth-wise Separable Convolutional Neural Networks (DWSC-NN) for image analysis.

Main Results:

• Achieved a high accuracy score of 94.16% in mask detection.
• Obtained an F1 score of 96.009%, indicating strong performance in identifying masked individuals.
• PCA effectively reduced irrelevant features, enhancing detection rates.

Conclusions:

• The proposed deep learning approach demonstrates significant effectiveness in detecting mask usage.
• The combination of PCA and DWSC-NN offers a robust solution for automated mask screening.
• This methodology contributes to public health safety by enabling reliable mask compliance monitoring.