Reproduce “Causal Spatiotemporal State Evolution for Monocular 4D Reconstruction” 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
Qiaowei Miao, Yijing Guo, Lipei Zhang, Qifeng Cheng, Yuru Xiao, Ziqian Ni, Sheng Yang
CSM reconstructs 3D geometry and motion from monocular video. It propagates geometry-aware representations in video order and uses UMS to combine geometry-induced flow with learned residual motion. On a controlled Sintel ablation, adding UMS lowers CSM's depth Abs Rel from 0.253 to 0.243 and camera ATE from 0.124 to 0.116, supporting gains in this setup without demonstrating eliminated temporal drift.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.