Reproduce “XM4D: 4D Reconstruction through Certifiable Geometric Optimization” 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
Haoyu Han, Jack Benarroch Jedlicki, Jiahui Lei, Heng Yang
XM4D reconstructs dynamic 3D scenes from monocular video by separating camera motion from scene deformation. It estimates cameras from static tracks, then models moving objects with an embedded deformation graph and certifiable geometric optimization before photometric refinement. On three rendering benchmarks, it has the best dataset-average scores for every reported metric; certificates concern optimization, not input correctness.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.