Reproduce “ReconDreamer#: Self-Evolving Closed-Loop Synergy of Reconstruction and Generation for Dynamic Driving Scenes” 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

Songling Liu, Guosheng Zhao, Xiaofeng Wang, Chaojun Ni, Guan Huang, Jiwen Lu, Xingang Wang, Zheng Zhu

ReconDreamer# uses self-evolution to train DriveRestorer on shifted-view failures: an auxiliary scene turns restored novel-trajectory views into degraded renders at recorded camera positions, paired with real images. Adaptive Ground Model constrains road support. On Waymo at a 6-m shift, scores improve over ReconDreamer++: NTA-IoU 0.489 to 0.532, NTL-IoU 56.57 to 58.29, and FID 111.92 to 94.99; unseen geometry remains unverified.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.