WIPUNet, a U-Net with residual subtraction, sigma maps, SE attention, and learned resampling, beats vanilla U-Net at high Gaussian noise by 0.3 to 1.2 dB.
FFDNet: Toward a Fast and Flexible Solution for CNN- based Image Denoising,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
WIPUNet: A Physics-inspired Network with Weighted Inductive Biases for Image Denoising
WIPUNet, a U-Net with residual subtraction, sigma maps, SE attention, and learned resampling, beats vanilla U-Net at high Gaussian noise by 0.3 to 1.2 dB.