Reproduce “MoRE-GS: Motion-Regime Experts for Future 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
Shima Salehi, Joshua Peeples
MoRE-GS forecasts future 3D scene views by combining motion predictors for soft keypoint groups and forecast horizons while keeping the reconstructed scene fixed. It calibrates routing per scene using observed frames. On an eight-scene D-NeRF test, it averages 27.47 dB PSNR, 0.9480 SSIM, and 0.0446 LPIPS, outperforming an ODE-GS reproduction in the mean, though not on every scene or metric.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.