3D Upper Body Reconstruction with Sparse Soft Sensors.

Area of Science:

• Computer Vision
• Biomedical Engineering
• Wearable Technology

Background:

• Traditional 3D body reconstruction systems often require restrictive clothing and raise privacy concerns.
• Existing methods struggle with user comfort and applicability in diverse environments.

Purpose of the Study:

• To develop a privacy-preserving and comfortable 3D upper body reconstruction system.
• To explore the use of soft sensors on standard garments for capturing body shape.
• To enable dynamic 3D reconstruction through a learning-based approach.

Main Methods:

• Utilizing a sparse array of soft sensors integrated into a standard garment.
• Modeling nonlinear sensor performance based on maximal stretching range.
• Analyzing the relationship between garment mesh deformation and sensor readings.
• Employing a learning-based approach for dynamic body shape reconstruction.

Main Results:

• The developed garment prototype is comfortable and wearable in various environments.
• Achieved state-of-the-art 3D reconstruction performance.
• Demonstrated significant advantages in privacy protection compared to vision-based systems.
• Validated through extensive testing with diverse users and over multiple days.

Conclusions:

• The soft sensor-based garment offers a practical and private solution for 3D upper body reconstruction.
• The system is suitable for long-term monitoring, including breath analysis.
• This approach expands the application scenarios for 3D body scanning technology.