Reproduce “SF3D-RGB: Scene Flow Estimation from Monocular Camera and Sparse LiDAR” 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
Rajai Alhimdiat, Ramy Battrawy, René Schuster, Didier Stricker, Wesam Ashour
SF3D-RGB estimates sparse 3D scene flow by extracting RGB features in the image domain, attaching them to calibrated LiDAR points, and matching fused point features with optimal transport before refining motion. On KITTId without fine-tuning, it achieves 0.227 m EPE3D versus 0.311 m for LiDAR-only FLOT at 2,048 sampled points, excluding ground; results concern sparse point motion, not dense reconstruction.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.