Reproduce “Reconstruct What You Can: Distractor-free 4D Gaussian Splatting” 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
Joel Julin, Liuyue Xie, Ananya Bal, Mose Sakashita, Koichiro Niinuma, Laszlo A. Jeni
ReCan-GS separates persistent dynamic scene content from transient distractors in monocular video by scoring variation in a shared deformation field and routing high-scoring content to per-view Gaussians or pruning it. On Bottle, it reaches 23.0 dB PSNR versus DeformableGS’s 21.6 dB on distractor-free holdout frames. Jeep ratings show lower distractor retention, but do not establish target completeness under occlusion.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.