Reproduce “CLEAR: Color-Texture Latent Alignment for Single Image Reflection Removal” 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

Tu Vo, Sheir A. Zaheer, Chan Y. Park

CLEAR removes reflections from photographs through glass by estimating transmission, reflection, and residue in a two-stage latent pipeline. HFAM then aligns input-derived high-frequency detail with clean-transmission features before decoding, rather than copying skips directly. Across six benchmarks and 1,040 images, CLEAR reports average PSNR 26.59, SSIM 0.906, and LPIPS 0.106, outperforming DAI on average but not in PSNR on Nature and Real20.

Summary by alphaXiv.

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