×
Perception-Oriented Single Image Super-Resolution using Optimal Objective Estimation. Single-image super-resolution (SISR) networks trained with perceptual and adversarial losses provide high-contrast outputs compared to those of networks trained with distortion-oriented losses, such as $L1$ or $L2$.
Nov 24, 2022 · The predictive model is trained using pairs of LR images and corresponding optimal objective maps searched from the objective trajectory.
Single-image super-resolution (SISR) networks trained with perceptual and adversarial losses provide high- contrast outputs compared to those of networks ...
The proposed SROOE shows a higher PSNR, LRPSNR, and lower LPIPS than other state-of-the-art methods, ie, lower distortion and higher perceptual quality.
Single-image super-resolution (SISR) networks trained with perceptual and adversarial losses provide high- contrast outputs compared to those of networks ...
Perception-oriented single image super-resolution (SISR), aiming to generate a high-resolution (HR) image with better perceptual quality from its ...
Perception-Oriented Single Image Super-Resolution using Optimal Objective Estimation (CVPR 2023) Link Seung Ho Park, Young Su Moon, Nam Ik Cho
People also ask
We provide more visual comparisons of 4× SISR be- tween our proposed SROOE and other state-of-the-art methods including a distortion-oriented method, ...
The current state-of-the-art on DIV2K val - 4x upscaling is SROOE. See a full comparison of 20 papers with code.
Perception-Oriented Single Image Super-Resolution using Optimal Objective Estimation · Flexible Style Image Super-Resolution using Conditional Objective.