Reproduce “DeFlow: Learning Complex Image Degradations from Unpaired Data with Conditional Flows” 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
Valentin Wolf, Andreas Lugmayr, Martin Danelljan, Luc Van Gool, Radu Timofte
DeFlow learns degradations from unpaired clean and degraded collections. It uses a shared invertible flow and Gaussian shift in latent space to generate varied degraded versions paired with clean inputs for super-resolution training. On AIM-RWSR, LPIPS is 0.349 versus 0.394 for Frequency Separation using the same super-resolution pipeline. This result is limited to that benchmark; unpaired inference remains ill-posed.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.