AI Paper Atlas

Research map / Label-efficient and robust learning / Test-time adaptation

Domain-adaptive Video Deblurring via Test-time Blurring

Jin-Ting He, Fu-Jen Tsai, Jia-Hao Wu, Yan-Tsung Peng, Chung-Chi Tsai, Chia-Wen Lin, Yen-Yu Lin

ECCV 2024 · Label-efficient and robust learning › Test-time adaptation · approach: Test-time adaptation of vision-language models · 6 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 DBCGM-generated blur conditions (with random patches) instead of random blur conditions (ablation)ESTRNN test-time adaptation · BSD-1ms8msPSNR: 23.88 -> 29.01Improved
use optical flows as blur conditions instead of DBCGM (ablation)ESTRNN test-time adaptation · BSD-1ms8msPSNR: 25.57 -> 25.51No clear effect
add RSDM sharp-patch selection over random patches (with DBCGM) (ablation)ESTRNN test-time adaptation · BSD-1ms8msPSNR: 29.01 -> 29.44Improved
add RSDM over random patches with random blur conditions (ablation)ESTRNN test-time adaptation · BSD-1ms8msPSNR: 23.88 -> 24.32Improved
random blur conditions with random patches fine-tuning (ablation)ESTRNN test-time adaptation · BSD-1ms8msPSNR: 25.57 -> 23.88Hurt
increase pseudo-sharp ratio r beyond 20% (ablation)RSDM · BSD-1ms8msPSNRHurt
use blurrier patches (excluding the sharpest) as pseudo-sharp (ablation)RSDM · BSD-1ms8msPSNRHurt
apply the proposed adaptation versus the meta-learning method of Liu et al.ESTRNN · BSD-3ms24msPSNR: 25.15 -> 28.32Improved
add test-time adaptation to four video deblurring modelsESTRNN, MMP-RNN, DSTNet, Shift-Net · five real-world deblurring datasetsPSNR: 21.63 -> 29.17 (MMP-RNN, BSD-1ms8ms)Improved

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