Reproduce “ClassSR: A General Framework to Accelerate Super-Resolution Networks by Data Characteristic” 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
Xiangtao Kong, Hengyuan Zhao, Yu Qiao, Chao Dong
ClassSR reduces computation in large-image super-resolution by routing image patches to versions of an existing network with different capacities. A learned classifier assigns patches by restoration difficulty, and the selected branches reconstruct and combine their outputs. On Test8K, ClassSR-FSRCNN reports 32.73 dB PSNR at 238M FLOPs, versus 32.66 dB at 468M for FSRCNN-O; this measures operations, not runtime.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.