AI summaryⓘ
The authors developed BioModule, a small add-on tool that can be attached to any 3D human pose estimation system to predict biomechanical information from standard 3D skeleton data. This is important because many fields need details about how the body moves and loads, not just where the key points are. They created a combined dataset linking videos and biomechanical labels, allowing BioModule to learn accurate biomechanical predictions. The authors also tested BioModule with several top pose estimators, showing how better pose data leads to better biomechanical results. This makes it easier to turn visual motion data into meaningful body movement analysis without changing existing pose estimation methods.
3D human pose estimationbiomechanical quantitiestemporal transformer17-joint 3D skeletonHuman3.6M datasetcross-modal supervisionpose estimation accuracybiomechanical predictionmarkerless motion capturemodular machine learning
Abstract
Recent progress in 3D human pose estimation has made markerless recovery of skeletal motion increasingly accurate and scalable. However, most pose estimators remain optimized for geometric keypoint accuracy, while many real-world applications in rehabilitation, sports science, ergonomics, and clinical movement analysis require biomechanical quantities that describe how the body moves, loads, and activates. In this work, we propose BioModule, a lightweight plug-in temporal transformer that attaches downstream of any 3D pose estimator and predicts biomechanical attributes from standard 17-joint 3D skeletons. BioModule is estimator-agnostic and requires no modification of the upstream pose model, enabling existing pose estimators to be extended toward physically interpretable motion analysis. To train and evaluate BioModule, we construct a large-scale aligned dataset pairing Human3.6M video and 3D keypoints with the biomechanical label space of Human3.6Mplus. We establish and verify anatomical correspondence between coordinate systems of the two datasets, enabling frame-accurate cross-modal supervision. Using this aligned supervision, BioModule predicts biomechanical quantities. We further benchmark BioModule across seven state-of-the-art 3D pose estimators, providing the first systematic analysis of how upstream pose estimation quality propagates to downstream biomechanical prediction fidelity. The results position BioModule as a compact, modular bridge between vision-based pose estimation and biomechanically meaningful human motion analysis.