Summary
Editor's Note: The following summary details independent academic research conducted in clinical research settings. Theia3D is an offline software solution engineered exclusively for research and human performance analysis.
Why This Matters
By removing physical markers and using AI-driven motion estimation, Theia3D enables a faster, less intrusive workflow while maintaining accuracy suitable for biomechanics research environments.
This study highlights how markerless technology can complement traditional motion capture, paving the way for more accessible, child-friendly gait evaluation.
Discover how Theia3D is being used in research settings.
Study Overview
Ten typically developing children participated in the LJMU study. Gait data were collected using a 10-camera marker-based system (Qualisys Oqus cameras) and a 7-camera markerless system (Qualisys Miqus cameras), both operating at 85 Hz.
Markerless data were processed through Theia3D’s deep-learning engine (v2023.1.0.3161) to estimate 3D joint and segment positions and orientations, while marker data (corresponding to the Plug-In Gait model) were labelled and gap-filled. Both datasets were imported into Visual3D, where a manually-defined model was applied to the marker data and an automatic model was generated based on the markerless data. The skeletal models were then compared across identical gait cycles.
Researchers used Generalized Additive Mixed Models (GAMMs) to account for time, age, and random variation across trials, providing a robust, statistical basis for comparing systems.
For a breakdown of how Theia3D captures 3D kinematics across real-world environments, see how it works.
Key Findings
- Comparable overall accuracy: Differences between Theia3D and PiG outputs were consistent with previous studies (typically <10°).
- Large offsets in pelvis and hip rotation: Some systematic differences were observed, likely reflecting differences in model definitions between methods.
- Low error in ankle and foot angles: Pelvic rotation and ankle dorsiflexion/plantarflexion differed by less than 5°, indicating strong agreement.
- Minor influence of age: Variability across age groups suggests gait maturity may affect model agreement, but no consistent pattern emerged.
What This Means for Pediatric Gait Biomechanics Research
For practitioners and researchers, these findings underscore how markerless motion capture can enhance feasibility and throughput in pediatric biomechanics. The system’s ability to generate robust data without markers reduces setup time and participant discomfort, especially valuable for young children or research participants with sensory sensitivities.
While Theia3D and PiG should not yet be used interchangeably for every research decision, this study confirms that markerless technology is a credible, complementary tool for pediatric gait biomechanics research and large-scale movement data collection.
Access the full peer-reviewed article here.