AI Paper Atlas

Research map / Image generation / Image restoration

Cross-Scale Pansharpening via ScaleFormer and the PanScale Benchmark

Ke Cao, Xuanhua He, Xueheng Li, Lingting Zhu, Yingying Wang, Ao Ma, Zhanjie Zhang, Man Zhou, Chengjun Xie, Jie Zhang

CVPR 2026 · Image generation › Image restoration · approach: Pansharpening and spectral fusion · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add Scale-Aware Patchify bucket sampling in training (ablation)ScaleFormer · Landsat-1600PSNR: 40.3895 -> 41.0335Improved
add RoPE to Sequence Transformers (ablation)ScaleFormer · Landsat-1600PSNR: 40.6924 -> 41.0335Improved
use Sequence Transformer instead of Spatial Transformer for inter-patch modeling (ablation)ScaleFormer · Landsat-1600PSNR: 40.5065 -> 41.0335Improved
use Sequence Transformer instead of Spatial Transformer for inter-patch modeling (ablation)ScaleFormer · Landsat-200PSNR: 40.9061 -> 40.6063Hurt
add RoPE to Sequence Transformers (ablation)ScaleFormer · Landsat-200PSNR: 40.4625 -> 40.6063Mixed
ScaleFormer vs ARConv with tiled inferenceScaleFormer · Skysat-AvePSNR: 43.4037 -> 44.6546Improved
ScaleFormer vs HFIN on full-resolution SkysatScaleFormer · Full-Skysat-AveQNR: 0.9403 -> 0.9408No clear effect
ScaleFormer vs HFIN on full-resolution JilinScaleFormer · Full-Jilin-AveQNR: 0.9155 -> 0.9156No clear effect
decoupled spatial/sequence modeling reduces parameters and compute versus ARConv/HFINScaleFormer · Efficiency at 200x200GFLOPs: 46.2067 -> 20.5679Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy