Reproduce “High-Resolution Image Inpainting using Multi-Scale Neural Patch Synthesis” 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

Chao Yang, Xin Lu, Zhe Lin, Eli Shechtman, Oliver Wang, Hao Li

The authors present an image-inpainting method combining global content prediction with local neural-feature patch matching. It uses VGG-19 features to match patches from known image regions, while a content constraint preserves structure across scales. On 128×128 Paris StreetView tests, it reports L1 loss 10.01%, L2 loss 2.21%, and PSNR 18.00 dB; pixel metrics do not measure realism.

Summary by alphaXiv.

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