AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

SAH-SCI: Self-Supervised Adapter for Efficient Hyperspectral Snapshot Compressive Imaging

Haijin Zeng, Yuxi Liu, Yongyong Chen, Youfa Liu, Chong Peng, Jingyong Su

ECCV 2024 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Pretext-task visual pre-training · 4 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
apply SAH self-supervised adapter to pre-trained reconstruction modelsHDNet · ICVLavg PSNR: 28.91 -> 34.78Improved
apply SAH adapter to all 7 pre-trained modelsend-to-end and unfolding models · ICVLavg PSNR: over 2dB average PSNR improvementImproved
apply SAH to DAUHST on NTIRE2022DAUHST · NTIRE2022PSNR: 38.44 -> 38.71No clear effect
add frequency adaption module to SAH (ablation)SAH on λ-Net · ICVLPSNR: 29.21 -> 29.95Improved
use 3-layer instead of 2-layer SAH (ablation)SAH on λ-Net · ICVLPSNR: 29.95 -> 30.15No clear effect
replace normal convolution with spatial-spectral convolution (ablation)SAH on λ-Net · ICVLPSNR: 29.95 -> 29.45 PSNR, params 2.09M -> 0.26MMixed
add channel weighting to SSConv (ablation)SAH on λ-Net · ICVLPSNR: 29.45 -> 29.98Improved
add equivariance imaging loss (ablation)SAH-SCI on HDNet · ICVLPSNR: 33.35 -> 34.31Improved
add image uncertainty loss (ablation)SAH-SCI on HDNet · ICVLPSNR: 33.35 -> 33.79Improved
add total variation loss (ablation)SAH-SCI on HDNet · ICVLPSNR: 34.50 -> 34.78Improved
full fine-tuning of pre-trained model instead of adapterHDNet · ICVL/Harvard/NTIREPSNR/SSIMHurt

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