Reproduce “Learning to Recorrupt: Noise Distribution Agnostic Self-Supervised Image Denoising” 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

Brayan Monroy, Jorge Bacca, Julián Tachella

L2R trains a denoiser and learned recorruptor together from noisy images without the exact noise distribution. Its selected structure reflects broad noise properties, and synthetic corruption discourages copying noise. Across non-Gaussian tests, it has the highest PSNR among distribution-agnostic methods in 11 of 12 settings, but not every comparison, including Laplace noise on DIV2K.

Summary by alphaXiv.

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