Are You Wearing a Mask? Detecting If a Person Wears a Mask Using a Wristband

Area of Science:

• Biomedical Engineering
• Public Health Technology
• Wearable Sensors

Background:

• The COVID-19 pandemic highlighted the need for effective public health interventions.
• Individual compliance with hygiene protocols, such as mask-wearing, is crucial for disease prevention.
• Monitoring adherence to these protocols can enhance public health strategies.

Purpose of the Study:

• To develop and evaluate an activity recognition method for detecting compliance with personal hygiene management codes.
• To assess the effectiveness of a wristband sensor equipped with an IR array and IMU for this purpose.
• To compare the performance of conventional machine learning algorithms and CNNs for activity recognition.

Main Methods:

• A wristband device incorporating an IR array and IMU was utilized.
• Data from 25 subjects performing various activities, including mask-wearing, were collected.
• Time series data from IMU and thermal images from the IR array were analyzed using machine learning and CNNs.

Main Results:

• The best performance was achieved using CNNs with 24 × 32 thermal images, reaching 97.8% accuracy.
• Similar high accuracy (97.1%) was obtained with low-resolution 6 × 8 thermal images.
• IMU data, while less accurate (93%), showed suitability for low-computational wearable devices.

Conclusions:

• The proposed wristband sensor system effectively recognizes activities related to personal hygiene compliance.
• IR array thermal imaging, particularly with CNNs, offers high accuracy for detecting behaviors like mask-wearing.
• The IMU component provides a viable, lower-power alternative for activity recognition in wearable technology.