Project R-15459

Assessing validity and reliability of the prototype: video-based markerless motion capture during walking and running. (Research)

The gold standard in the assessment of normal and pathologic human movement is laboratory-based optoelectronic three-dimensional motion analysis. Unfortunately, the need to attach markers and the need for a high-tech laboratory inhibit the routine collection of valuable high quality data, because it is very expensive and time-consuming and requires technical expertise. Furthermore, motion data collected in a laboratory may fail to capture how individuals move in natural settings. There are very few affordable systems able to accurately analyze movement (e.g., walking and running) in clinical/rehabilitation and sport settings. Recent advances in biomedical engineering resulted in new techniques based on deep learning to track body landmarks in simple video recordings, which include a high degree of automatization, and allow recordings in an unobtrusive manner in a natural environment. The manually placed, skin-mounted markers could be replaced with automatically detected landmarks with these deep learning based techniques. The development and application of markerless systems that can estimate human posture under a variety of conditions is increasing. Many researchers are conducting validation and/or reliability studies, comparing a markerless tool with 3D marker-based motion capture. Until now, no markerless tool has been validated and is reliable to measure joint kinematics in all planes for both the upper body and the lower body during walking and running AND is user-friendly for non-movement analysts or engineers (in clinical and/or sport settings). Therefore, we are developing a user-friendly tool to accurately analyze movement in simple video recordings in clinical and sport settings

16 January 2025 - 15 January 2026