Reproduce “MuyBridge: Mobile Human Center-of-Mass Estimation from Monocular Video via Sparse Fusion” 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
Aidan Bradshaw, Marco Giordano, David Rode, Andreas Habersack, Elif Basokur, Annika Kruse, Markus Tilp, Michele Magno and 3 more
MuyBridge, a mobile, single-camera system, estimates human center-of-mass (CoM) trajectories in real-time on devices such as an iPhone 15. The system achieves vertical CoM errors between 33 mm and 41 mm across diverse movements by sparsely fusing monocular 2D pose and depth information with anatomical and physical priors.
Key insights
- Efficient on-device CoM estimation is achievable by combining optimized monocular 2D pose and depth with an analytic metric fusion framework.
- Metric scale for CoM can be robustly recovered from affine-invariant monocular depth by integrating sparse anthropometric and physical cues.
- Extensive network compression, restructuring, and quantization strategies enable complex computer vision pipelines for pose and depth to run in real-time on resource-constrained mobile devices.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.