Reproduce “BenthicFlow: Generating Extensible Underwater Environments via Flow Matching” 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
Joaquín Figueira, Camile Lendering, Manfred Gonzalez-Hernandez, Giacomo D'Amicantonio, Erkut Akdag, Egor Bondarev
BenthicFlow generates aligned color-depth imagery from underwater survey cues and enlarges it by averaging overlapping windows’ flow predictions during one shared trajectory. It lifts estimated depth into renderable surfel scenes. An overlap sweep found seam-gradient ratios closest to 1 at 62.5% overlap, while reef-classification scores retained site-specific visual cues. The output remains a single-view 2.5D representation, not recovered unseen geometry.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.