AI Paper Atlas

Research map / Image generation / Image restoration

Practical Compact Deep Compressed Sensing

Bin Chen, Jian Zhang

TPAMI 2024 · Image generation › Image restoration · approach: Inverse imaging with learned priors · 43 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 transmission augmenting (31-channel auxiliary variable across stages) (ablation)PGD-unrolled baseline · Set11PSNR (dB): +0.58dB PSNRImproved
enable ratio scalability with cropped Gaussian matrix (ablation)PGD-unrolled baseline · Set11PSNR (dB): -0.31dB PSNRHurt
make sampling matrix learnable (ablation)PGD-unrolled baseline · Set11PSNR (dB): +1.97dB PSNRImproved
replace residual blocks with Swin-Conv blocks (ablation)PGD-unrolled baseline · Set11PSNR (dB): +0.26dB PSNR, -0.26M paramsImproved
add spatial downscaling r=2 in Swin-Conv blocks (ablation)PGD-unrolled baseline · Set11PSNR (dB): -0.06dB PSNR, more than 2x speedMixed
train on larger WED dataset instead of Train400 (ablation)PGD-unrolled baseline · Set11PSNR (dB): +0.23dB PSNRImproved
use Charbonnier loss (ablation)PGD-unrolled baseline · Set11PSNR (dB): +0.07dB PSNRImproved
add deep filtering network to sampling operator (ablation)PC-CNN · Set11 CS ratio 10% (PC-CNN ablation)PSNR (dB): 30.42 -> 31.23Improved
add D-branch (global DCT) to G-branch sampling (ablation)PC-CNN · Set11 CS ratio 10% (PC-CNN ablation)PSNR (dB): 29.97 -> 30.42Improved
add random permutation to G-branch (ablation)PC-CNN · Set11 CS ratio 10% (PC-CNN ablation)PSNR (dB): 29.31 -> 30.42Improved
add G-branch to D-branch (ablation)PC-CNN · Set11 CS ratio 10% (PC-CNN ablation)PSNR (dB): 29.43 -> 30.42Improved
increase stage number K from 10 to 20 (ablation)PC-CNN · Set11 CS ratio 10% (PC-CNN ablation)PSNR (dB): 31.06 -> 31.39Improved

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