High-Resolution Be Aware! Improving the Self-Supervised Real-World Super-Resolution

Angela Yao; Yuehan Zhang

REVIEW 5 major objections 4 minor 42 references

High-Resolution Be Aware! Improving the Self-Supervised Real-World Super-Resolution

T0 review · 5 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A quality gauge and a feature-alignment regularizer let self-supervised super-resolution adapt to real-world domains without paired data.

desk verdict Plausible components undermined by a transductive evaluation that uses HR labels from the test set to pick the model; the method may work, but Table 1 doesn't show it. read the letter →

arxiv 2411.16175 v1 pith:U4LAYK3F submitted 2024-11-25 eess.IV cs.CV

Yuehan Zhang , Angela Yao This is my paper · ORCID

classification eess.IVcs.CV

keywords self-supervisedsuper-resolutionreal-worlddegradationembeddingcontrollerfeature-alignmentregularizerLPIPSCLIPtest-timeadaptationperceptualquality

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper argues that self-supervised super-resolution underperforms because it relies on degradation models built from low-resolution inputs alone, missing what high-resolution natural images look like. It proposes two additions: a controller that scales the degradation embedding according to how close the super-resolved output is to a bicubic upsampling of the input, and a feature-alignment regularizer that pulls the statistics of super-resolved images toward those of natural high-resolution images as judged by a fixed CLIP encoder. Finetuning off-the-shelf SR models on real-world test sets with these components lowers perceptual distance and raises no-reference quality scores while keeping fidelity roughly stable. If correct, this gives a practical recipe to adapt any pretrained SR model to a specific camera or domain without collecting paired high-resolution data.

What carries the argument

The central mechanism is the low-resolution reconstruction network from the LWay paradigm—a degradation encoder, an image encoder, and a reconstructor—augmented by two new parts. The first is a scalar controller $s$ that multiplies the 512-dimensional degradation embedding $\mathbf{e}_d$, defined as $s = n + (1 - \mathrm{HQI})$ in pretraining and $s = n + \mathrm{HQI}$ in finetuning, where $n$ is Gaussian noise and $\mathrm{HQI} = 1 - \mathrm{LPIPS}(f_\uparrow(X^r), M(X^r))$ measures how close the super-resolved output is to a bicubic upsampling of the low-resolution input. The second is the Feature-Alignment Regularizer (FAR), which computes average- and max-pooled Gram-matrix statistics of features from a trainable image encoder and a frozen CLIP encoder, aligns their dimensions with linear layers, and penalizes the Frobenius-norm difference. FAR is trained on natural high-resolution images during pretraining, then applied to super-resolved outputs during finetuning, pulling the output distribution toward natural-image statistics.

What would settle it

Construct a synthetic dataset where low-resolution images are degraded by independent and unequal amounts of blur and noise, then finetune an off-the-shelf SR model with the controller and FAR. If two images with identical HQI but different blur-noise mixes cause the controller to produce higher LPIPS after finetuning than the off-the-shelf model, or if sweeping the scalar $s$ while holding all else fixed changes reconstruction error along an unintended degradation axis, the scalar-control assumption fails.

Watch

Extended reading notes

Core claim

The paper establishes that the low-resolution reconstruction objective can be realigned with super-resolution quality by (i) multiplying the degradation embedding with a scalar controller that tracks the LPIPS distance between the super-resolved output and the bicubic-upsampled low-resolution input, and (ii) adding a feature-alignment regularizer that enforces the Gram-matrix statistics of the super-resolved output to match those of natural high-resolution images as encoded by a frozen CLIP image encoder. The controller prevents the SR model from degrading its own output to satisfy the reconstruction loss, while the regularizer constrains the output distribution toward natural-image statistics that are invisible in the low-resolution input alone. On NTIRE20, AIM2019, and RealSR benchmarks, finetuning RealESRGAN+ and SwinIRGAN with these components yields lower LPIPS and higher NRQM than prior self-supervised finetuning baselines, with comparable or better fidelity.

Load-bearing premise

The method assumes that a single scalar—how similar the super-resolved output is to the bicubic upsampling of the low-resolution input—correctly orders images by degradation extent, and that multiplying the whole degradation embedding by this scalar changes only the amount of degradation without altering its type.

Editorial extensions

If this is right

Any off-the-shelf SR model, GAN-based or transformer-based, can be adapted to a target real-world domain by finetuning with the reconstruction loss plus the quality-scaled controller and FAR, requiring no paired high-resolution data.
FAR constrains the distribution of super-resolved outputs toward natural-image statistics, which should reduce over-smoothing and unnatural textures that pure reconstruction objectives produce, as reflected in higher NRQM and lower LPIPS.
The controller's dependence on HQI makes the finetuning self-regulating: if the SR model degenerates toward bicubic interpolation, the controller increases the degradation load, preventing the model from trivializing the reconstruction objective.
Finetuning becomes more stable over many iterations: the paper shows LPIPS does not degrade after 600 iterations with FAR, whereas the LWay baseline overfits the reconstruction objective and suffers performance drops.
The method operates on a whole test set with consistent degradation, making it suitable for camera-specific adaptation rather than per-image optimization.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The HQI based on LPIPS ties the quality measure to the bicubic upsampling of the low-resolution input; a no-reference quality metric that does not depend on that reference could generalize to settings where bicubic upsampling is a poor baseline, such as heavily compressed low-resolution images.
FAR uses CLIP as the fixed encoder, which biases the distribution constraint toward semantic features; ablations show that an SR-trained encoder (EDSR) gives higher NRQM but worse fidelity, suggesting a multi-encoder ensemble might better balance perceptual quality and fidelity.
The controller scales the entire degradation embedding by a single scalar; a per-degradation-type adjustment (blur vs noise vs compression) could be more effective on real-world data where these degradations are independent, as the paper itself lists finer degradation adjustments as a future direction.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Two things to know before you read this paper. The two proposed mechanisms—the quality-controlled degradation embedding (controller s) and the CLIP-based feature-alignment regularizer (FAR)—are sensible, and the ablations suggest each does something on the metrics tested. But the headline comparison in Table 1 is not a self-supervised evaluation. The authors finetune on the full test set and select the stopping iteration using LPIPS on a validation split of that same test set. Baselines SRTTA and the off-the-shelf models are not given the same HR-guided selection. So the reported LPIPS and NRQM gains are largely explained by selecting models with ground-truth HR information from the test set, not by the method's own self-supervision.

What is actually new: grafting a scalar quality-scaled degradation controller and a CLIP feature-alignment regularizer onto the LWay reconstruction paradigm is a legitimate extension, though not a new framework. The controller design is heuristic—one scalar multiplier on a 512-dimensional embedding—but the paper gives a nice diagnostic (Fig. 2b) showing why plain LR reconstruction can favor degraded outputs. FAR is also plausible, and Table 2 supports that it helps perceptual metrics.

Soft spots, in proportion: the evaluation protocol is the load-bearing flaw. Section 5.2 states they split a validation set from the testing data and use early stopping for LWay and Ours; Fig. 9 shows they pick the iteration with best LPIPS on those validation images. Since the test set is also the finetuning set, the model is selected by peeking at HR labels. The paper's robustness claim in Sec. 5.3 (that the method is appropriate when early stopping is infeasible) is not supported by any table—all Table 1 numbers use the HR-guided selection. There are also minor issues: no error bars, no significance tests, no code release, and LWay is an unofficial reproduction. The scalar controller assumption is heuristic but not fatal; the ablation confirms it helps.

Bottom line: the method is worth a look, but the claimed state-of-the-art perceptual quality is not evidenced. A corrected evaluation with a truly held-out validation set or no early stopping is needed. This paper is for people working on test-time adaptation for SR, and as a cautionary example of transductive evaluation pitfalls. I would send it to peer review, but with the requirement that the authors rerun without HR-guided early stopping and report both settings.

Referee Report

5 major / 4 minor

Summary. The manuscript proposes two additions to the LR-reconstruction self-supervised SR framework of LWay: a controller that rescales the degradation embedding by an LPIPS-based quality indicator (HQI), and FAR, a Gram-matrix feature-alignment regularizer that encourages super-resolved outputs to match CLIP feature statistics of natural HR images. The method is applied by finetuning pretrained RealESRGAN+ and SwinIRGAN on target real-world LR test sets. Reported results on NTIRE20, AIM2019, and RealSR show improved LPIPS and NRQM over SRTTA, LWay, and the off-the-shelf models, with ablations attributing the gains to both components.

Significance. The two components are simple, intuitive, and clearly described, and the ablation structure is a sensible way to isolate their contributions. A notable strength is that the authors explicitly disclose the finetuning and early-stopping protocol, which allows the reader to assess the evidence. If the reported gains survive a clean evaluation protocol, the controller and FAR would be useful additions to test-time self-supervised SR. However, the central evidence is currently protocol-dependent, the LWay baseline is an unofficial reproduction, no code is released, and no error bars are reported, so the state-of-the-art claim is not yet established.

major comments (5)

[Sec. 5.2, Sec. 5.3, Table 1, Fig. 9] The headline evaluations select models using ground-truth HR information in a transductive protocol. Sec. 5.2 states that finetuning runs on the whole testing set and that for LWay and Ours 'we split a validation set from the testing data and use the early stopping strategy'; Fig. 9 then selects stopping iterations by LPIPS on those validation images. Thus the final models in Table 1 are chosen with access to HR references on images drawn from the same test sets, while the SRTTA and off-the-shelf baselines are not given the same HR-guided selection. This makes the reported improvements unattainable from LR-only self-supervision and breaks the comparability of Table 1. Please provide results under a clean protocol (fixed iteration count chosen without HR, or a validation split from an independent domain) and report both early-stopped and non-early-stopped numbers.
[Sec. 5.2] The only comparison to LWay uses an unofficial reproduction because LWay does not release source code or training data. Since Table 1's state-of-the-art claim depends on outperforming LWay, an uncontrolled reproduction can bias the comparison. Please release the reproduction details (training iterations, reconstruction-loss weights, and the exact validation schedule) and ideally use the official implementation or provide a verification check, such as reproducing LWay's published numbers on a common benchmark.
[Tables 1-4] No repeated runs, error bars, or significance tests are reported. Several headline margins are small (e.g., NTIRE20 with RealESRGAN+: LPIPS 0.2369 vs. 0.2498 for LWay; NRQM 6.137 vs. 6.121 for the off-the-shelf model). Without seed variance or a paired test, the improvements cannot be distinguished from noise. Please report mean plus/minus standard deviation over at least three seeds and apply a paired significance test for the main comparisons.
[Eqs. (7)-(8), Sec. 4.3] The controller's sign and the HQI ordering are assumed rather than derived. In Eq. (8), the pretraining and finetuning definitions use opposite signs of HQI, and the finetuning relation is justified only by the observation that a smaller LR/SR distance indicates more degradation in the SR output. Because LPIPS is also the primary evaluation metric, the objective and the metric are entangled; improvements in LPIPS may reflect optimization of the same distance rather than genuine perceptual quality. The paper should include a sensitivity analysis of the controller sign and evaluate with perceptual metrics not used in the loss (e.g., DISTS, NIQE, FID, or a human study). The NRQM gains mitigate this concern but do not eliminate it.
[Table 2] The ablation table does not specify the evaluation dataset, and its baseline numbers (PSNR 26.2, SSIM 0.7889, LPIPS 0.2525, NRQM 4.511) do not match any off-the-shelf model row in Table 1 for NTIRE20, AIM2019, RealSR-Canon, or RealSR-Nikon. Please state the dataset and settings used for Table 2 and explain the discrepancy.

minor comments (4)

[Eq. (7), Table 1, Fig. 2, Conclusion] There are several typos: 'cosince' should be 'cosine' in Eq. (7); 'KernalGAN' should be 'KernelGAN' in Table 1; the Fig. 2 caption contains 'inX' instead of 'in X'; and the conclusion uses 'fintunes' instead of 'finetunes'.
[Sec. 5.3, LR reconstruction vs. SR performance] The interpolation experiment in Eq. (12) and Fig. 7 does not state which dataset is used, nor whether the controller s is applied during pretraining, finetuning, or both. Please add these details so the experiment can be reproduced.
[General] No code availability statement is provided. Releasing the code and pretrained reconstruction network would substantially improve reproducibility, especially since the LWay baseline must be reimplemented.
[Sec. 4.4, reference [30]] The motivation for FAR relies on reference [30]'s observation that self-supervised tasks degrade under distribution shift, but the connection between that observation and Gram-matrix alignment with CLIP is informal. A short explanation of why this particular alignment should reveal distribution shift would help readers evaluate the design.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the derivation chain is empirical and self-contained, with a protocol caveat that is not circular under the rubric.

full rationale

The paper's derivation chain does not reduce its claimed contributions to its own inputs. The controller s in Eq. (8) is defined through the external LPIPS-based HQI of Eq. (7), and the FAR regularizer in Eq. (11) aligns features to the external CLIP encoder; neither is defined in terms of the reported evaluation numbers. The reported LPIPS and NRQM gains are aligned with the training objectives by design, but that is objective alignment, not a circular derivation: no equation makes the headline metric equal to a fitted parameter or to the method's own definition. The only self-citation is reference [40], used as a related-work pointer to unsupervised SR methods, and it is not load-bearing for any argument. One non-circular caveat is worth flagging: Sec. 5.2 states that finetuning runs on the whole testing set and that for LWay and Ours a validation set is split from the testing data for early stopping (Fig. 9); this makes Table 1 transductive and weakens generalization claims, but it is an evaluation-protocol concern, not a circularity of the derivation. The core methods are empirically evaluated against external benchmarks and pretrained models, so the central claim has independent content.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

No new physical or ontological entities are introduced. The controller and HQI are functions of existing quantities (LPIPS, degradation embeddings, CLIP features). The axioms listed are the main unproved premises the method depends on.

free parameters (3)

FAR loss weights λpt and λft = not reported
Balance reconstruction loss against feature-alignment regularization; the values are chosen by validation but not stated in the main text.
Finetuning iterations and early-stopping criterion = 300-600 iterations; early stopping on validation split from test set
Controls overfitting of the LR reconstruction objective; the validation split comes from the same test set used for evaluation, making the selection transductive.
Controller scaling sign for HQI in finetuning = s = n + HQI in finetuning
Table 4 compares s = n + HQI with s = n + 1 - HQI and selects the better variant, i.e., a design choice fit to validation performance.

assumptions (4)

domain assumption A well-trained self-supervised task indicates distribution shift by performance drop, so FAR loss values can reveal whether SR outputs depart from natural HR distribution.
Sec 4.1 and Fig. 8; based on an observation from [30], not proven for CLIP features on SR outputs.
domain assumption CLIP image encoder features on natural HR images define a valid target manifold for super-resolved images.
Sec 4.4; no evidence that CLIP features discriminate natural HR statistics at the patch level relevant to SR.
domain assumption LPIPS distance between bicubic-upsampled LR and SR output is a monotone proxy for SR quality that can drive the controller.
Eq. (7), Sec 4.3; high-quality SR outputs differ from LR by construction, so using low distance as low quality is a heuristic.
domain assumption The degradation embedding from a network pretrained on synthetic second-order degradation (DIV2K) generalizes to real-world degradations and remains valid when scalar-multiplied by the controller.
Sec 3.2 and Eq. (4); inherited from LWay; scalar scaling of the embedding is an added assumption.

how reviews work

0 comments

Cite this review

Pith. "Pith review of High-Resolution Be Aware! Improving the Self-Supervised Real-World Super-Resolution." pith.science (2026). https://pith.science/paper/U4LAYK3F

@misc{pith2026241116175,
  author       = {Pith},
  title        = {Pith review of: High-Resolution Be Aware! Improving the Self-Supervised Real-World Super-Resolution},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/U4LAYK3F}},
  note         = {Machine review of arXiv:2411.16175}
}

read the original abstract

Self-supervised learning is crucial for super-resolution because ground-truth images are usually unavailable for real-world settings. Existing methods derive self-supervision from low-resolution images by creating pseudo-pairs or by enforcing a low-resolution reconstruction objective. These methods struggle with insufficient modeling of real-world degradations and the lack of knowledge about high-resolution imagery, resulting in unnatural super-resolved results. This paper strengthens awareness of the high-resolution image to improve the self-supervised real-world super-resolution. We propose a controller to adjust the degradation modeling based on the quality of super-resolution results. We also introduce a novel feature-alignment regularizer that directly constrains the distribution of super-resolved images. Our method finetunes the off-the-shelf SR models for a target real-world domain. Experiments show that it produces natural super-resolved images with state-of-the-art perceptual performance.

Figures

Figures reproduced from arXiv: 2411.16175 by the authors.

**Figure 2.** (a) Architecture of LR reconstruction network in [ [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** Overview of our method. In each stage, only colored modules are optimized, and LR images are interpolated for better visuals. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (5 more)

**Figure 4.** Figure 4: Details of the Feature-Alignment Regularizer. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

**Figure 5.** Figure 5: Visual comparison with state-of-the-art methods (zoom-in for better views). RE refers to RealESRGAN+ and columns (e-g) are [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

**Figure 6.** Figure 6: Visual results for DPED [10] dataset. Methods in (c-d) finetune SwinIRGAN (Swin). LWay fails to improve the oversmoothness produced by Swin. Our method reconstructs realistic textures that better resemble the original images. 5.3. Ablations Effectiveness of Designs. T…

**Figure 7.** Figure 7: Plot of LR reconstruction performance changing with [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]

**Figure 8.** Figure 8: Histogram of FAR loss values on natural HR images [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 29 canonical work pages

[1]

Ntire 2017 challenge on single image super-resolution: Dataset and study

Eirikur Agustsson and Radu Timofte. Ntire 2017 challenge on single image super-resolution: Dataset and study. In Pro- ceedings of the IEEE conference on computer vision and pat- tern recognition workshops, pages 126–135, 2017. 6

2017
[2]

Toward real-world single image super-resolution: A new benchmark and a new model

Jianrui Cai, Hui Zeng, Hongwei Yong, Zisheng Cao, and Lei Zhang. Toward real-world single image super-resolution: A new benchmark and a new model. In Proceedings of the IEEE/CVF international conference on computer vision, pages 3086–3095, 2019. 2, 6

work page 2019
[3]

Real-world blind super-resolution via feature matching with implicit high- resolution priors

Chaofeng Chen, Xinyu Shi, Yipeng Qin, Xiaoming Li, Xi- aoguang Han, Tao Yang, and Shihui Guo. Real-world blind super-resolution via feature matching with implicit high- resolution priors. In Proceedings of the 30th ACM Interna- tional Conference on Multimedia , pages 1329–1338, 2022. 2

work page 2022
[4]

Low-res leads the way: Improving generalization for super- resolution by self-supervised learning

Haoyu Chen, Wenbo Li, Jinjin Gu, Jingjing Ren, Haoze Sun, Xueyi Zou, Zhensong Zhang, Youliang Yan, and Lei Zhu. Low-res leads the way: Improving generalization for super- resolution by self-supervised learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 25857–25867, 2024. 1, 2, 3, 5, 6, 7, 8

work page 2024
[5]

Zero-shot im- age super-resolution with depth guided internal degradation learning

Xi Cheng, Zhenyong Fu, and Jian Yang. Zero-shot im- age super-resolution with depth guided internal degradation learning. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceed- ings, Part XVII 16, pages 265–280. Springer, 2020. 1, 3

work page 2020
[6]

Efficient test-time adapta- tion for super-resolution with second-order degradation and reconstruction

Zeshuai Deng, Zhuokun Chen, Shuaicheng Niu, Thomas Li, Bohan Zhuang, and Mingkui Tan. Efficient test-time adapta- tion for super-resolution with second-order degradation and reconstruction. Advances in Neural Information Processing Systems, 36:74671–74701, 2023. 1, 2, 3, 6, 7

work page 2023
[7]

Im- age style transfer using convolutional neural networks

Leon A Gatys, Alexander S Ecker, and Matthias Bethge. Im- age style transfer using convolutional neural networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 2414–2423, 2016. 5

work page 2016
[8]

Blind super-resolution with iterative kernel correction

Jinjin Gu, Hannan Lu, Wangmeng Zuo, and Chao Dong. Blind super-resolution with iterative kernel correction. In Proceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition, pages 1604–1613, 2019. 1, 2, 3, 6

work page 2019

Show all 42 references

[9]

Un- folding the alternating optimization for blind super resolu- tion

Yan Huang, Shang Li, Liang Wang, Tieniu Tan, et al. Un- folding the alternating optimization for blind super resolu- tion. Advances in Neural Information Processing Systems , 33:5632–5643, 2020. 2

2020
[10]

Dslr-quality photos on mobile devices with deep convolutional networks

Andrey Ignatov, Nikolay Kobyshev, Radu Timofte, Kenneth Vanhoey, and Luc Van Gool. Dslr-quality photos on mobile devices with deep convolutional networks. InProceedings of the IEEE international conference on computer vision, pages 3277–3285, 2017. 7

2017
[11]

Analyzing and improv- ing the image quality of stylegan

Tero Karras, Samuli Laine, Miika Aittala, Janne Hellsten, Jaakko Lehtinen, and Timo Aila. Analyzing and improv- ing the image quality of stylegan. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 8110–8119, 2020. 6

2020
[12]

Adam: A method for stochastic opti- mization

Diederik P Kingma. Adam: A method for stochastic opti- mization. arXiv preprint arXiv:1412.6980, 2014. 6

2014 arXiv
[13]

Photo- realistic single image super-resolution using a generative ad- versarial network

Christian Ledig, Lucas Theis, Ferenc Husz´ar, Jose Caballero, Andrew Cunningham, Alejandro Acosta, Andrew Aitken, Alykhan Tejani, Johannes Totz, Zehan Wang, et al. Photo- realistic single image super-resolution using a generative ad- versarial network. In Proceedings of the IE...
[14]

Swinir: Image restoration us- ing swin transformer

Jingyun Liang, Jiezhang Cao, Guolei Sun, Kai Zhang, Luc Van Gool, and Radu Timofte. Swinir: Image restoration us- ing swin transformer. InProceedings of the IEEE/CVF inter- national conference on computer vision , pages 1833–1844,
[15]

Enhanced deep residual networks for single image super-resolution

Bee Lim, Sanghyun Son, Heewon Kim, Seungjun Nah, and Kyoung Mu Lee. Enhanced deep residual networks for single image super-resolution. In Proceedings of the IEEE confer- ence on computer vision and pattern recognition workshops, pages 136–144, 2017. 8

2017
[16]

Catch missing details: Image reconstruction with frequency augmented variational autoencoder

Xinmiao Lin, Yikang Li, Jenhao Hsiao, Chiuman Ho, and Yu Kong. Catch missing details: Image reconstruction with frequency augmented variational autoencoder. In Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1736–1745, 2023. 2

2023
[17]

On bayesian adaptive video super resolution

Ce Liu and Deqing Sun. On bayesian adaptive video super resolution. IEEE transactions on pattern analysis and ma- chine intelligence, 36(2):346–360, 2013. 2

2013
[18]

Degae: A new pretraining paradigm for low-level vision

Yihao Liu, Jingwen He, Jinjin Gu, Xiangtao Kong, Yu Qiao, and Chao Dong. Degae: A new pretraining paradigm for low-level vision. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition , pages 23292–23303, 2023. 2

2023
[19]

Sgdr: Stochas- tic gradient descent with warm restarts

Ilya Loshchilov and Frank Hutter. Sgdr: Stochas- tic gradient descent with warm restarts. arXiv preprint arXiv:1608.03983, 2016. 6

2016 arXiv
[20]

Aim 2019 challenge on real-world image super-resolution: Methods and results

Andreas Lugmayr, Martin Danelljan, Radu Timofte, Manuel Fritsche, Shuhang Gu, Kuldeep Purohit, Praveen Kandula, Maitreya Suin, AN Rajagoapalan, Nam Hyung Joon, et al. Aim 2019 challenge on real-world image super-resolution: Methods and results. In 2019 IEEE/CVF International C...

2019
[21]

Ntire 2020 challenge on real-world image super-resolution: Methods and results

Andreas Lugmayr, Martin Danelljan, and Radu Timofte. Ntire 2020 challenge on real-world image super-resolution: Methods and results. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition Work- shops, pages 494–495, 2020. 6

2020
[22]

Learning a no-reference quality metric for single-image super-resolution

Chao Ma, Chih-Yuan Yang, Xiaokang Yang, and Ming- Hsuan Yang. Learning a no-reference quality metric for single-image super-resolution. Computer Vision and Image Understanding, 158:1–16, 2017. 6

2017
[23]

Learning transferable visual models from natural language supervi- sion

Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervi- sion. In International conference on machine learning, ...

2021
[24]

Unpaired real-world super-resolution with pseudo controllable restora- 9 tion

Andr ´es Romero, Luc Van Gool, and Radu Timofte. Unpaired real-world super-resolution with pseudo controllable restora- 9 tion. In Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition, pages 798–807, 2022. 2

2022
[25]

zero-shot

Assaf Shocher, Nadav Cohen, and Michal Irani. “zero-shot” super-resolution using deep internal learning. In Proceed- ings of the IEEE conference on computer vision and pattern recognition, pages 3118–3126, 2018. 1, 2, 3, 6

2018
[26]

Very deep convo- lutional networks for large-scale image recognition

Karen Simonyan and Andrew Zisserman. Very deep convo- lutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556, 2014. 8

2014 arXiv
[27]

Meta- transfer learning for zero-shot super-resolution

Jae Woong Soh, Sunwoo Cho, and Nam Ik Cho. Meta- transfer learning for zero-shot super-resolution. In Proceed- ings of the IEEE/CVF conference on computer vision and pattern recognition, pages 3516–3525, 2020. 1, 2, 3

2020
[28]

A+: Adjusted anchored neighborhood regression for fast super- resolution

Radu Timofte, Vincent De Smet, and Luc Van Gool. A+: Adjusted anchored neighborhood regression for fast super- resolution. In Computer Vision–ACCV 2014: 12th Asian Conference on Computer Vision, Singapore, Singapore, November 1-5, 2014, Revised Selected Papers, Part IV 12 , pa...

2014
[29]

Ntire 2018 chal- lenge on single image super-resolution: Methods and results

Radu Timofte, Shuhang Gu, and Jiqing Wu. Ntire 2018 chal- lenge on single image super-resolution: Methods and results. In 2018 IEEE/CVF Conference on Computer Vision and Pat- tern Recognition Workshops (CVPRW) , pages 965–96511,

2018
[30]

Convolu- tional visual prompt for robust visual perception

Yun-Yun Tsai, Chengzhi Mao, and Junfeng Yang. Convolu- tional visual prompt for robust visual perception. Advances in Neural Information Processing Systems, 36:27897–27921,
[31]

Exploiting diffusion prior for real-world image super-resolution

Jianyi Wang, Zongsheng Yue, Shangchen Zhou, Kelvin CK Chan, and Chen Change Loy. Exploiting diffusion prior for real-world image super-resolution. International Journal of Computer Vision, pages 1–21, 2024. 2

2024
[32]

Real-esrgan: Training real-world blind super-resolution with pure synthetic data

Xintao Wang, Liangbin Xie, Chao Dong, and Ying Shan. Real-esrgan: Training real-world blind super-resolution with pure synthetic data. In Proceedings of the IEEE/CVF inter- national conference on computer vision , pages 1905–1914,

1905
[33]

Unsupervised real-world image super resolution via domain-distance aware training

Yunxuan Wei, Shuhang Gu, Yawei Li, Radu Timofte, Long- cun Jin, and Hengjie Song. Unsupervised real-world image super resolution via domain-distance aware training. In Pro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 13385–13394, 2021. 2

2021
[34]

Image super-resolution as sparse representation of raw image patches

Jianchao Yang, John Wright, Thomas Huang, and Yi Ma. Image super-resolution as sparse representation of raw image patches. In 2008 IEEE conference on computer vision and pattern recognition, pages 1–8. IEEE, 2008. 2

2008
[35]

Deep unfold- ing network for image super-resolution

Kai Zhang, Luc Van Gool, and Radu Timofte. Deep unfold- ing network for image super-resolution. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 3217–3226, 2020. 2

2020
[36]

Designing a practical degradation model for deep blind image super-resolution

Kai Zhang, Jingyun Liang, Luc Van Gool, and Radu Timo- fte. Designing a practical degradation model for deep blind image super-resolution. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 4791– 4800, 2021. 2

2021
[37]

The unreasonable effectiveness of deep features as a perceptual metric

Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shecht- man, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. In Proceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 586–595, 2018. 5, 6

2018
[38]

Crafting training degradation distribution for the accuracy-generalization trade-off in real- world super-resolution

Ruofan Zhang, Jinjin Gu, Haoyu Chen, Chao Dong, Yulun Zhang, and Wenming Yang. Crafting training degradation distribution for the accuracy-generalization trade-off in real- world super-resolution. In International conference on ma- chine learning, pages 41078–41091. PMLR, 2023. 2, 3

2023
[39]

Real-world image super-resolution as multi-task learning

Wenlong Zhang, Xiaohui Li, Guangyuan Shi, Xiangyu Chen, Yu Qiao, Xiaoyun Zhang, Xiao-Ming Wu, and Chao Dong. Real-world image super-resolution as multi-task learning. Advances in Neural Information Processing Systems , 36,
[40]

Pairwise dis- tance distillation for unsupervised real-world image super- resolution

Yuehan Zhang, Seungjun Lee, and Angela Yao. Pairwise dis- tance distillation for unsupervised real-world image super- resolution. arXiv preprint arXiv:2407.07302, 2024. 2

2024 arXiv
[41]

Joint sub-bands learning with clique structures for wavelet domain super-resolution

Zhisheng Zhong, Tiancheng Shen, Yibo Yang, Zhouchen Lin, and Chao Zhang. Joint sub-bands learning with clique structures for wavelet domain super-resolution. Advances in neural information processing systems, 31, 2018. 2

2018
[42]

Guided frequency separation network for real-world super- resolution

Yuanbo Zhou, Wei Deng, Tong Tong, and Qinquan Gao. Guided frequency separation network for real-world super- resolution. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, pages 428–429, 2020. 2 10

2020

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

Ntire 2017 challenge on single image super-resolution: Dataset and study

Eirikur Agustsson and Radu Timofte. Ntire 2017 challenge on single image super-resolution: Dataset and study. In Pro- ceedings of the IEEE conference on computer vision and pat- tern recognition workshops, pages 126–135, 2017. 6

2017

[2] [2]

Toward real-world single image super-resolution: A new benchmark and a new model

Jianrui Cai, Hui Zeng, Hongwei Yong, Zisheng Cao, and Lei Zhang. Toward real-world single image super-resolution: A new benchmark and a new model. In Proceedings of the IEEE/CVF international conference on computer vision, pages 3086–3095, 2019. 2, 6

work page 2019

[3] [3]

Real-world blind super-resolution via feature matching with implicit high- resolution priors

Chaofeng Chen, Xinyu Shi, Yipeng Qin, Xiaoming Li, Xi- aoguang Han, Tao Yang, and Shihui Guo. Real-world blind super-resolution via feature matching with implicit high- resolution priors. In Proceedings of the 30th ACM Interna- tional Conference on Multimedia , pages 1329–1338, 2022. 2

work page 2022

[4] [4]

Low-res leads the way: Improving generalization for super- resolution by self-supervised learning

Haoyu Chen, Wenbo Li, Jinjin Gu, Jingjing Ren, Haoze Sun, Xueyi Zou, Zhensong Zhang, Youliang Yan, and Lei Zhu. Low-res leads the way: Improving generalization for super- resolution by self-supervised learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 25857–25867, 2024. 1, 2, 3, 5, 6, 7, 8

work page 2024

[5] [5]

Zero-shot im- age super-resolution with depth guided internal degradation learning

Xi Cheng, Zhenyong Fu, and Jian Yang. Zero-shot im- age super-resolution with depth guided internal degradation learning. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceed- ings, Part XVII 16, pages 265–280. Springer, 2020. 1, 3

work page 2020

[6] [6]

Efficient test-time adapta- tion for super-resolution with second-order degradation and reconstruction

Zeshuai Deng, Zhuokun Chen, Shuaicheng Niu, Thomas Li, Bohan Zhuang, and Mingkui Tan. Efficient test-time adapta- tion for super-resolution with second-order degradation and reconstruction. Advances in Neural Information Processing Systems, 36:74671–74701, 2023. 1, 2, 3, 6, 7

work page 2023

[7] [7]

Im- age style transfer using convolutional neural networks

Leon A Gatys, Alexander S Ecker, and Matthias Bethge. Im- age style transfer using convolutional neural networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 2414–2423, 2016. 5

work page 2016

[8] [8]

Blind super-resolution with iterative kernel correction

Jinjin Gu, Hannan Lu, Wangmeng Zuo, and Chao Dong. Blind super-resolution with iterative kernel correction. In Proceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition, pages 1604–1613, 2019. 1, 2, 3, 6

work page 2019