Reproduce “ReSMap: Recasting Satellite Priors for Robust and Accurate Online HD Map Construction” 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
Kyungmin Kim, Sumin Lee, Sungoh Jeong, DoHyun Lim, Seonghyun Park, Soonmin Hwang
ReSMap uses cached satellite imagery as a backup source for online HD mapping when vehicle cameras are unreliable. It estimates confidence on bird’s-eye-view grids to weight camera and satellite features, then uses learned gates to merge modality-specific updates for map elements. On a geographically non-overlapping nuScenes split, it retains 40.1 mAP under all-camera blackout, versus 47.3 clean; this test does not establish satellite imagery alone caused the retained score.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.