Reproduce “Frame-Recurrent Video Super-Resolution” 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
Mehdi S. M. Sajjadi, Raviteja Vemulapalli, Matthew Brown
FRVSR performs video super-resolution by carrying the previous high-resolution estimate forward, aligning it to the next low-resolution frame, and combining it with that frame to predict the next estimate. On Vid4 at 4×, FRVSR 10-128 reports 26.69 dB PSNR and 0.822 SSIM, versus 26.16 dB and 0.815 for Bayesian SR under the paper’s evaluation protocol.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.