Reproduce “Unaligned Image Guided Denoising via Cross-modal Conditional 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
Runmin Zhang, Linshan Li, Si-Yuan Cao, Zhu Yu, Lingyu Zhu, Huaqi Zhang, Hui-Liang Shen
UGD-FM denoises noisy RGB images with unaligned near-infrared (NIR) guidance. It progressively restores RGB while matching guidance, using sparse propagation and a two-step flow-matching solver. On FANVID with Poisson–Gaussian noise, its Restormer variant reaches 30.58 dB PSNR, versus 28.92 dB for SANet. This is one result among gains reported across tested datasets and noise settings; severe parallax can still make guidance unreliable.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.