Reproduce “Hihaze: One-Step Diffusion for High-Resolution Real-World Image Dehazing” 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
Zheng Chen, Yuchen Huang, Xiaoyang Liu, Jue Gong, Linghe Kong, Yulun Zhang
Hihaze restores high-resolution hazy photographs with a one-step diffusion model, using local haze guidance and attentive skip compensation to adapt restoration. Global-guided patch fusion aligns overlapping tile predictions to a full-image appearance reference while retaining local detail. Across four real-world benchmarks without fine-tuning, it reports 17.29 dB PSNR on Dense-Haze, versus 13.67 for PTTD, though color discrepancies remain.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.