Reproduce “From Skin to Skeleton: Towards Biomechanically Accurate 3D Digital Humans” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Marilyn Keller, Keenon Werling, Soyong Shin, Scott Delp, Sergi Pujades, C. Karen Liu, Michael J. Black
Keller et al. introduce SKEL, a 3D digital human model that integrates a biomechanically accurate skeleton with a plausible body surface, allowing for realistic posing and compatibility with existing data-driven models. The work develops the BioAMASS dataset to learn the relationship between skin and internal skeletal structures, demonstrating improved anatomical joint prediction and enabling the enhancement of existing pose and shape datasets with biomechanical parameters.
Key insights
- It is feasible to predict biomechanically accurate internal skeletal joint locations and bone orientations from the external surface of parametric body models through data-driven regression.
- Incorporating realistic joint models for complex articulations like the spine, shoulder blades, and forearms significantly improves the anatomical plausibility and functional constraints of digital human models.
- A unified model can bridge computer vision/graphics and biomechanics by combining the data-driven flexibility and surface accuracy of models like SMPL with the anatomical precision of biomechanical skeletons.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.