Self-Improved Privilege Learning uses a learned dictionary to let an all-in-one restoration model refine its own output at test time, reporting PSNR gains up to +4.58 dB on composite degradations.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Boosting All-in-One Image Restoration via Self-Improved Privilege Learning
Self-Improved Privilege Learning uses a learned dictionary to let an all-in-one restoration model refine its own output at test time, reporting PSNR gains up to +4.58 dB on composite degradations.