Author Identifier
Marion Mundt: https://orcid.org/0000-0001-6624-2895
Document Type
Journal Article
Publication Title
Scientific Reports
Volume
15
Issue
1
PubMed ID
40281018
Publisher
Nature
School
Nutrition and Health Innovation Research Institute
Funders
Centre for the Analysis of Motion, Entertainment Research and Applications (EP/M023281/1, EP/T014865/1) / Mobility Grant of the International Society of Biomechanics in Sport
Abstract
Markerless motion capture has the potential to enable biomechanical analyses without specialised, high-cost equipment. However, the comparability of many markerless motion capture frameworks with the most used marker-based method is limited. One reason for this is the lack of high-quality, biomechanically-informed datasets that are needed to train markerless models. This study aimed to inform the development of such a dataset by systematically analysing the agreement between a gold-standard marker set and a reduced number of markers to solve inverse kinematics (IK). We analysed the impact of different marker positions on the IK solution using an OpenSim lower body model with real and synthetic data of running, walking and counter movement jumps. We found that one mid-segment marker in addition to two anatomical markers per segment result in the best agreement to a gold-standard marker set. The results for real and synthetic data across all movements were similar, with synthetic data showing slightly better agreement with a reduced number of markers (root mean squared error 1.55–8.27° real data, 1.27–7.79° synthetic data), likely due to limited soft tissue artefacts and missing human error in marker placement. These findings can support the development of a dataset to retrain markerless models incorporating biomechanical knowledge.
DOI
10.1038/s41598-025-97219-5
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Comments
Mundt, M., Pagnon, D., & Colyer, S. (2025). The influence of the marker set on inverse kinematics results to inform markerless motion capture annotations. Scientific Reports, 15, 14547. https://doi.org/10.1038/s41598-025-97219-5