Lower limb pose estimation for monitoring the kicking patterns of infants

Area of Science:

• Biomedical Engineering
• Computer Vision
• Developmental Pediatrics

Background:

• Infant spontaneous kicking patterns offer valuable insights into neurodevelopment.
• Accurate kinematic analysis requires precise tracking of joint trajectories.
• Existing methods may lack the precision or accessibility for infant movement studies.

Purpose of the Study:

• To develop and validate a computer vision-based method for estimating infant leg pose from range images.
• To enable quantitative analysis of infant kicking kinematics.
• To provide a tool for monitoring developmental changes in motor patterns.

Main Methods:

• Utilizing range imaging to capture infant leg data.
• Employing Robust Point Set Registration (RPSR) for articulated model fitting.
• Extracting joint trajectories over time from image sequences.

Main Results:

• The method successfully estimated joint trajectories for a robotic humanoid mimicking infant kicking.
• The technique was applied to real infant data, generating kinematic signal estimates.
• Validation demonstrated the feasibility of the approach for infant movement analysis.

Conclusions:

• The proposed computer vision method offers a viable approach for analyzing infant kicking kinematics.
• This technique can aid in the non-invasive monitoring of infant motor development.
• Further application can lead to earlier identification of developmental deviations.