Reproduce “Enhancing RAW-to-sRGB with Decoupled Style Structure in Fourier Domain” 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
Xuanhua He, Tao Hu, Guoli Wang, Zejin Wang, Run Wang, Qian Zhang, Keyu Yan, Ziyi Chen and 4 more
FourierISP converts phone RAW images into DSLR-like RGB by learning structure and style/color separately in the Fourier domain, then combining them. It treats phase as structure and amplitude as style/color, without assuming they are fully independent. On MAI, after ZRR pretraining and one epoch of fine-tuning, it reports 24.99 dB PSNR versus 24.34 dB for MWISP-GAN, under adaptation rather than zero-shot transfer.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.