Reproduce “From Sentinel-2 to 1-m Imagery: Real-World $\times 10$ Cross-Sensor Super-Resolution via Visual Autoregressive Modeling” 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

Tailai Chen, Yuluxin Fu, Jie Deng, Shu Luo, Yanan Wu, Danfeng Hong

CrossVAR reconstructs 1-m full-scene imagery from a single natural-color Sentinel-2 composite. It predicts latent maps from coarse to fine, refines quantization residuals, and assembles aligned overlapping windows. On 3,015 YREB-SR test scenes, it reaches 27.6008 dB PSNR, above Thera’s 25.2978 dB. These scores measure agreement with cross-sensor references, not verified locations of sub-10-m features.

Summary by alphaXiv.

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