Reproduce “A Cosine Network for Image 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
Chunwei Tian, Chengyuan Zhang, Bob Zhang, Zhiwu Li, C. L. Philip Chen, David Zhang
CSRNet performs single-image super-resolution by alternating odd enhancement blocks with asymmetric parallel and serial convolutions, and even blocks with ordinary convolutions and skip connections. The authors train it using cosine learning-rate restarts. It reports the best PSNR and SSIM across four standard benchmark datasets and three scales, including 26.92 dB and 0.8098 on Urban100 at ×4; these results do not establish real-world robustness.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.