A Low-Resolution Infrared Array for Unobtrusive Human Activity Recognition That Preserves Privacy

Area of Science:

• Computer Science
• Sensor Technology
• Human-Computer Interaction

Background:

• Traditional human activity recognition often compromises user privacy.
• Camera-based systems raise significant privacy concerns.
• Existing methods may require complex machine learning models.

Purpose of the Study:

• To develop a real-time human activity recognition system that preserves privacy.
• To utilize low-resolution infrared array sensors for unobtrusive monitoring.
• To avoid machine learning for a more direct and efficient approach.

Main Methods:

• Employing a low-resolution infrared array sensor to capture thermal data.
• Representing captured thermal pixels as human silhouettes.
• Utilizing interpolation and mathematical measures for activity recognition.

Main Results:

• The system accurately recognizes multiple human states in real-time.
• Performance was validated against traditional machine learning approaches in various scenarios.
• The method demonstrated effectiveness without compromising user privacy.

Conclusions:

• Low-resolution infrared array sensors offer a viable solution for privacy-preserving activity recognition.
• The proposed unobtrusive method is suitable for applications in homes and healthcare settings.
• This approach provides an efficient and effective alternative to privacy-invasive methods.