Pith. sign in

REVIEW 4 major objections 6 minor 112 references

Augmenting Perceptual Super-Resolution via Image Quality Predictors

T0 review · 4 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read An automatic no-reference image-quality measure can replace human raters in super-resolution fine-tuning, matching or beating human-feedback training without any manual labels.

desk verdict A worthwhile empirical study of using NR-IQA to guide SR training, with an honest limitations section; the main claim is plausible but the circularity between the optimized metric and the headline evaluation metric means it should be read with that caveat. read the letter →

arxiv 2504.18524 v1 pith:JCMIP7B6 submitted 2025-04-25 cs.CV

Fengjia Zhang , Samrudhdhi B. Rangrej , Tristan Aumentado-Armstrong , Afsaneh Fazly , Alex Levinshtein This is my paper · ORCID
classification cs.CV
keywords single-imagesuper-resolutionno-referenceimagequalityassessmentperceptualhumanfeedbackMUSIQLoRAfine-tuningperception-distortiontradeoffmulti-ground-truthsupervision
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 a pretrained no-reference image-quality assessment (NR-IQA) model can stand in for human preference feedback when training single-image super-resolution (SISR) models. On two human-labeled SR datasets, the authors test which NR-IQA metrics align best with human judgments and select MUSIQ as the most reliable. They then use MUSIQ in two ways: to reweight which of several candidate high-resolution targets the network is trained on, and as a differentiable objective in a LoRA-regularized fine-tuning step. The combined method, called AMO+FT, is evaluated on the held-out HGGT Test-100 benchmark, where it reports higher perceptual-quality scores than the human-feedback baseline on all four NR-IQA metrics tested and was preferred by human raters 69.7% of the time in a 12-rater study. If correct, the practical payoff is that perceptually strong super-resolution can be trained without expensive manual annotation.

What carries the argument

The central object is MUSIQ (Multi-Scale Image Quality Transformer), a learned no-reference image quality model, used twice: as a scoring function for choosing training targets and as a differentiable loss for fine-tuning. The paper first analyses 42 NR-IQA variants to justify selecting MUSIQ for its human alignment and low positive-misalignment rate. In sampling, MUSIQ weights a softmax distribution over candidate ground truths, and in the 'argmax-online' (AMO) variant it selects the highest-quality patch among several super-resolved candidates at training time. In fine-tuning, adding $\lambda_Q Q(\hat{I})$ to the loss would normally act as an adversarial attack on the quality network, so the paper restricts updates to low-rank adaptation (LoRA) parameters, which removes the structured artifacts while still letting the model improve its quality score.

What would settle it

A content-diverse user study in which human raters consistently prefer the human-feedback baseline over AMO+FT outputs, or prefer images with lower MUSIQ scores, would refute the central claim; a sharper test is to search for SR image pairs with similar MUSIQ scores where one is visibly artifact-laden, since such pairs would show the score is not a dependable quality proxy.

Watch

Extended reading notes

Core claim

The paper's central claim is that pretrained NR-IQA models can act as a fully automatic substitute for human rankings in perceptual super-resolution, and that combining two uses of them—quality-weighted sampling of multiple ground truths and regularized direct optimization—beats the human-feedback baseline while requiring no human annotations. On the held-out HGGT Test-100 set, the AMO+FT combination raises the MUSIQ score from 66.39 to 70.81 for SwinIR and from 65.93 to 71.67 for RealESRGAN, improves all four NR metrics over the UPos baseline, and wins the preference study with p<0.01, while taking only small penalties on mid-level reference metrics such as LPIPS and DISTS and larger losses on pixel-level PSNR and SSIM. The paper interprets this as moving to a more human-centric point on the perception-distortion tradeoff, prioritizing standalone image quality over pixelwise fidelity.

Load-bearing premise

The load-bearing premise is that MUSIQ's score reliably tracks human perceptual quality on super-resolved images, a premise the paper's own Limitations section concedes is not guaranteed, so that maximizing it improves perceived quality rather than merely gaming the metric.

Editorial extensions

If this is right

  • Super-resolution training can proceed without any human ranking of candidate ground truths; the automatic NR-IQA signal is sufficient to match or exceed the human-feedback baseline on perceptual metrics.
  • The perception-distortion tradeoff shifts: NR-IQA fine-tuning lowers pixel-level PSNR and SSIM while raising high-level perceptual scores, but unlike naive optimization it preserves or improves shift-tolerant mid-level similarity (LPIPS-ST).
  • The combination of online argmax sampling and LoRA-regularized fine-tuning gives the largest gains, more than either component alone, on both SwinIR and RealESRGAN architectures.
  • Which NR-IQA model is chosen matters: replacing MUSIQ with PaQ-2-PiQ degrades complementary perceptual metrics, so selecting a human-aligned metric is part of the recipe.
  • A 12-rater user study prefers the automatic AMO+FT outputs over the human-guided UPos baseline 69.7% of the time, with the rater mean significantly above 50% (p<0.01).

Reading between the lines

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

  • Beyond the paper, the reported NR-IQA gains are partly self-referential because MUSIQ is both the training signal and the lead evaluation metric; the paper's own Limitations section concedes that higher IQA scores do not guarantee improved human perceptual quality, so the complementary metrics and the user study carry extra weight.
  • Beyond the paper, the same sampling-plus-optimization recipe could transfer to other restoration tasks with subjective quality targets, such as denoising, deblurring, or compression-artifact removal, since no paired ground truth is needed at deployment.
  • Beyond the paper, a testable extension is to train or fine-tune an NR-IQA model specifically on super-resolution artifacts, or to combine multiple IQA signals, which the paper suggests could reduce the structured artifacts seen under naive optimization.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper studies whether no-reference image quality assessment (NR-IQA) models can replace human feedback in super-resolution training. It first evaluates a large set of NR-IQA metrics on two human-preference SR datasets (SBS180K and HGGT), identifying MUSIQ as the most aligned metric. It then proposes two mechanisms: (i) reweighted sampling of multiple ground-truth targets based on IQA scores (SMA, SMP, and AMO in §4.2), and (ii) fine-tuning with a differentiable NR-IQA loss regularized by LoRA (§4.3, Eq. 5). The combination AMO+FT is reported to outperform the human-feedback-based HGGT 'UPos' baseline on NR-IQA metrics and in a 12-participant user study (Supp. §12), supporting the claim that automatic IQA can match or exceed human-guided fine-tuning for SR.

Significance. If established, the result would be significant: it would show that modern NR-IQA metrics can act as a scalable automatic reward for perceptual SR optimization, avoiding costly human annotation. The paper is also valuable for its systematic comparison of 42 NR-IQA variants on human SR preference data, for the additional RealSR generalization results, and for the honest ablation showing that direct IQA optimization naively produces adversarial artifacts (Fig. 3) while LoRA mitigates them. However, the central 'perceptually on par or better than human feedback' claim is not yet airtight, because MUSIQ is simultaneously the optimization target and the headline evaluation metric, and the only human check covers one architecture with 12 raters.

major comments (4)
  1. [Table 5, Eq. (5), §4.2] The largest reported gains are on MUSIQ itself, which is both the training signal and the primary evaluation metric: for SwinIR-AMO+FT vs SwinIR-UPos, MUSIQ increases by 4.42 points while NIMA, Q-Align, and TOPIQ increase by 0.13, 0.19, and 0.08, respectively. Those non-optimized metrics are the only direct evidence of perceptual generalization, and their small magnitude is not discussed with significance or confidence intervals. Please report variances or error bars across test images, and add an evaluation metric that was never used during training and is not a monotone transform of MUSIQ, to break the circularity.
  2. [§5.1, Table 5, Gold Standard row] SwinIR-AMO+FT reaches MUSIQ 70.81, above the 'Gold Standard' value of 69.64 computed by taking the best MUSIQ score among the original and enhanced GTs per quintuplet. Since the fine-tuning loss maximizes MUSIQ directly, surpassing the GT bound is exactly what metric overfitting (Goodharting) would predict; the manuscript currently presents this as evidence of success without addressing the alternative explanation. Please add a human evaluation or an analysis of the learned artifacts that supports the interpretation that the model truly produces higher-quality images than the best available GT.
  3. [Supp. §12] The user study is too narrow to carry the general claim: it uses only 12 raters, only the SwinIR architecture, only the AMO+FT vs UPos comparison, and no comparison against the gold-standard GTs or the original HR images. In particular, there is no human validation for RealESRGAN, no test of whether preference is driven by the same artifacts flagged in Supp. §14 and Fig. 8, and no analysis of whether the 69.7% preference correlates with the MUSIQ margin. A larger, architecture-diverse human study with artifact-focused instructions is needed before claiming 'perceptually on par or better than SISR finetuned with human feedback'.
  4. [Supp. §14, Fig. 8] The paper's own limitations concede that higher IQA scores do not guarantee improved human perceptual quality and show aliasing-like line artifacts and mangled text under AMO+FT. Combined with the known susceptibility of NR-IQA models to adversarial perturbations (cited in §4.3), this means the possibility that AMO+FT partially games MUSIQ while shifting the sharpness/artifact tradeoff has not been excluded. Please add an artifact-level analysis (e.g., side-by-side with original GT, text-specific SR metrics, or a targeted human study) to rule this out.
minor comments (6)
  1. [§4.2] Clarify the temperature convention in SoftMax_τ: with the standard softmax, τ→0 yields argmax and τ→∞ uniform, matching the text, but the notation is easy to misread; please define the formula explicitly, e.g., SoftMax_τ(Q) = exp(Q/τ)/Σ exp(Q/τ).
  2. [§5.1] In the RealESRGAN paragraph, fix the typo 'neglible' and the inconsistent expansion 'DIST' vs 'DISTS'.
  3. [Table 5 caption] Add the ranges and directions of the NR-IQA metrics to the main-table caption (they are currently only in Supp. Table 7), so that absolute values like MUSIQ 70.81 vs 69.64 are interpretable.
  4. [Supp. §7.1] The statement that negatives are rare (∼6%) and MUSIQ ranks a negative highest in ∼34% of such cases is useful; consider moving it to the main text, since it is part of the justification for using MUSIQ despite poor NM in Table 4.
  5. [Supp. §8.3] Please provide any rationale or sensitivity analysis for the LoRA rank choices (rank 48 for SwinIR, 24 for RealESRGAN), since these could affect the artifact behavior observed in Fig. 8.
  6. [Eq. (5)] State explicitly that the negative sign before λ_Q Q(I) reflects that Q is higher-is-better, so the loss is minimized by increasing Q.

Circularity Check

2 steps flagged · score 6.0 of 10

MUSIQ is simultaneously the training objective, the sampling signal, and the headline evaluation metric, so the largest reported gains are partly forced by construction; independent NR metrics and a small user study provide only partial support.

  1. fitted input called prediction [§4.3 Eq. (5); §5 'Evaluation'; Table 5]
    "we continue training as normal, but only on the LoRA weights, plus an additional NR-IQA loss term: eL(ϕ|bI,I ) =L(ϕ|bI,I )−λQQ(bI) ... we use MUSIQ [41] as our NR IQA model, Q, for both weighted sampling and direct optimization."

    In Table 5, MUSIQ is the first 'NR High-Lev. Perceptual Quality' metric and the paper's main evidence of perceptual gains (e.g., SwinIR-AMO+FT reaches 70.81 vs UPos 66.39). But Q in Eq. 5 is set to MUSIQ, so the fine-tuned model is explicitly trained to maximize the same score being reported. The MUSIQ improvement is therefore a direct consequence of the optimization objective, not an independent confirmation that perceptual quality improved. The paper itself notes that naive MUSIQ optimization produces adversarial structured artifacts (§4.3, Fig. 3, Supp. §10.2). The other NR metrics (NIMA, Q-Align, TOPIQ) improve less, and the only human check is a 12-rater, SwinIR-only preference study, so the headline 'perceptually on par or better' is not disentangled from metric overfitting.

  2. other [§5.1 'Upper-bounding NR-IQA Evaluation Performance'; Table 5 'Gold Standard' row]
    "we compute 'gold standard' NR-IQA values for the GT Test data, taking the best score among the original and enhanced images, providing a soft upper-bound on NR scores, if one were able to exactly reproduce the 'best' GT via the SR network. We find that altered sampling generally does not reach these values, but FT is able to reach and even surpass them in several scenarios."

    The 'gold standard' is defined as the maximum MUSIQ score among the GT images. Since AMO+FT is trained with an explicit −λQ Q(bI) MUSIQ term, exceeding 69.64 is what Goodharting predicts: the optimizer can push the exact scalar above the best GT value while producing artifacts (as the paper's own Limitations and Fig. 3/Supp. §10.2 show). Using this bound as evidence of super-human quality is circular, because the bound and the training objective are the same function.

full rationale

The paper has genuinely independent components: the §3 NR-IQA analysis uses external human-preference datasets (SBS180K and HGGT) to select and justify MUSIQ, and the user study in Supp. §12 is a human check not based on MUSIQ. Those parts are not circular. However, the central empirical claim in §5 is partially circular because Q=MUSIQ appears in Eq. 2 (AMO sampling), Eq. 5 (fine-tuning objective), and Table 5 (the headline NR evaluation metric). The largest numerical improvements are on MUSIQ itself, which is the exact function being optimized; the 'gold standard' bound is also a MUSIQ maximum, so surpassing it is predicted by optimizing the same scalar. Independent evidence (NIMA, Q-Align, TOPIQ, the 12-rater user study) shows smaller or mixed gains, and the Limitations section explicitly concedes that higher IQA scores do not guarantee improved human perceptual quality. Thus the claim that automatic IQA yields results 'perceptually on par or better' than human-feedback fine-tuning is not fully established independently of the optimized metric. Score is 6 rather than higher because the user study and cross-metric improvements do exist, and the §3 metric selection uses external human judgments rather than the paper's own outputs.

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

The central claim depends on several hand-chosen hyperparameters and on the domain assumption that MUSIQ is a trustworthy perceptual oracle. No new entities are introduced.

free parameters (4)
  • Sampling temperature tau = 10
    Chosen by hand in Supp. §8.1 for SMA and SMP to interpolate between uniform and argmax sampling.
  • Fine-tuning loss weight lambda_Q = 0.05
    Set in Supp. §8.3 for the MUSIQ fine-tuning loss; no sensitivity analysis is shown.
  • LoRA ranks = 48 (SwinIR MLPs), 24 (RealESRGAN RRDBs)
    Chosen architecturally in Supp. §8.3; affects the capacity and hence the regularization of fine-tuning.
  • Fine-tuning steps = 20,000
    Fixed in Supp. §8.3; stated to yield rapid convergence but no ablation of this value is reported.
assumptions (3)
  • domain assumption MUSIQ's differentiable score is a useful and safe training signal when optimization is constrained to LoRA parameters.
    Invoked in §4.3 to justify adding -lambda_Q Q(I_hat) to the loss; the paper shows naive optimization causes artifacts, so the LoRA constraint is assumed sufficient. Supp. §14 concedes higher IQA score does not guarantee human quality.
  • domain assumption Human annotations in HGGT provide a valid set of positive ground truths, and the positives-only scenario is the best-performing HGGT variant.
    Relied on in §4.1 and §5 to define supervision sets and baselines; the paper reproduces HGGT results but does not independently validate the human labels.
  • domain assumption NR-IQA metrics evaluated on SBS180K and HGGT accurately reflect human preference in the SR setting.
    Used in §3 to select MUSIQ; Table 4 shows very low Spearman correlations (SRC at most 0.28), so this premise is partially contradicted by the paper's own data.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Augmenting Perceptual Super-Resolution via Image Quality Predictors." pith.science (2026). https://pith.science/paper/JCMIP7B6

@misc{pith2026250418524,
  author       = {Pith},
  title        = {Pith review of: Augmenting Perceptual Super-Resolution via Image Quality Predictors},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JCMIP7B6}},
  note         = {Machine review of arXiv:2504.18524}
}
read the original abstract

Super-resolution (SR), a classical inverse problem in computer vision, is inherently ill-posed, inducing a distribution of plausible solutions for every input. However, the desired result is not simply the expectation of this distribution, which is the blurry image obtained by minimizing pixelwise error, but rather the sample with the highest image quality. A variety of techniques, from perceptual metrics to adversarial losses, are employed to this end. In this work, we explore an alternative: utilizing powerful non-reference image quality assessment (NR-IQA) models in the SR context. We begin with a comprehensive analysis of NR-IQA metrics on human-derived SR data, identifying both the accuracy (human alignment) and complementarity of different metrics. Then, we explore two methods of applying NR-IQA models to SR learning: (i) altering data sampling, by building on an existing multi-ground-truth SR framework, and (ii) directly optimizing a differentiable quality score. Our results demonstrate a more human-centric perception-distortion tradeoff, focusing less on non-perceptual pixel-wise distortion, instead improving the balance between perceptual fidelity and human-tuned NR-IQA measures.

Figures

Figures reproduced from arXiv: 2504.18524 by the authors.

Figure 1
Figure 1. Schematics for improving perceptual super-resolution (SR). Perceptual quality of SR can be improved in two ways: (Left) providing supervision through multiple enhanced ground-truths (EGT) or (Right) direct optimization for the quality of the super-resolved image. In both cases, human-in-the-loop can greatly improve performance. However, manual annotation is tedious, imprecise, and non￾differentiable. An IQA metric c… view at source ↗
Figure 2
Figure 2. Fine-Grained Comparison via NR-IQA. MUSIQ can differentiate the quality of two images, both marked as “positive” by human annotators. Higher MUSIQ indicates higher quality (zoom for details). Unlike HGGT models, which utilize a uni￾form distribution over positives, our approach enables differently weighting them (§4.2). worse than the original). The set of positives provides mul￾timodal supervision, since each one i… view at source ↗
Figure 3
Figure 3. Structured Optimization Noise. Optimizing via an NR-IQA metric (MUSIQ [41]) generates structured artifacts (left), similar to an adversarial attack, while utilizing LoRA removes this noise (right; see §4.3 and Supp. §10.2). Zoom in for details. τ → 0). We eschew human data; hence, SI = AI . This enables a more fine-grained judgment (since quality is com￾puted at the patch level), whereas human annotations cannot nec… view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Qualitative results with NR-IQA Guidance. Following the notation of [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Structured Noise due to naive NR-IQA optimization. The left three insets show an image and two close-ups that was fine-tuned without LoRA, whereas the right three show the effect of using LoRA. Note the patterns that form in the sky and the strangely coloured pixels th…
Figure 6
Figure 6. Figure 6: Qualitative results with NR-IQA guidance. Following the notation of [PITH_FULL_IMAGE:figures/full_fig_p017_6.png]
Figure 7
Figure 7. Figure 7: User study example. Users can move the slider to alter￾nate between 2 images. mean: 4.8%), suggesting our algorithm is preferred over the HGGT-based UPos approach at a more than 2:1 ratio, de￾spite their use of human annotations, which ours does not use. Following simi…
Figure 8
Figure 8. Figure 8: Illustration of limitations. We show examples of short￾comings of our method (see [PITH_FULL_IMAGE:figures/full_fig_p019_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

112 extracted references · 73 canonical work pages

  1. [1]

    ARNIQA: Learning distortion manifold for image quality assessment

    Lorenzo Agnolucci, Leonardo Galteri, Marco Bertini, and Alberto Del Bimbo. ARNIQA: Learning distortion manifold for image quality assessment. In Winter Conference on Applications of Computer Vi- sion (WACV), 2024. 3, 5

    2024

  2. [2]

    Fast, accurate, and lightweight super- resolution with cascading residual network

    Namhyuk Ahn, Byungkon Kang, and Kyung-Ah Sohn. Fast, accurate, and lightweight super- resolution with cascading residual network. In Euro- pean Conference on Computer Vision (ECCV), 2018. 2

    2018

  3. [3]

    Explorable super resolution

    Yuval Bahat and Tomer Michaeli. Explorable super resolution. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2020. 2

    2020

  4. [4]

    Limits on super- resolution and how to break them

    Simon Baker and Takeo Kanade. Limits on super- resolution and how to break them. IEEE Transac- tions on Pattern Analysis and Machine Intelligence (PAMI), 2002. 1

    2002

  5. [5]

    The perception- distortion tradeoff

    Yochai Blau and Tomer Michaeli. The perception- distortion tradeoff. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018. 2, 8, 16, 18

    2018

  6. [6]

    Rethinking lossy compression: The rate-distortion-perception trade- off

    Yochai Blau and Tomer Michaeli. Rethinking lossy compression: The rate-distortion-perception trade- off. In International Conference on Machine Learn- ing (ICML), 2019. 2, 8

    2019

  7. [7]

    The 2018 PIRM challenge on perceptual image super-resolution

    Yochai Blau, Roey Mechrez, Radu Timofte, Tomer Michaeli, and Lihi Zelnik-Manor. The 2018 PIRM challenge on perceptual image super-resolution. In European Conference on Computer Vision Work- shops (ECCVW), 2018. 3

    2018

  8. [8]

    Deep neural networks for no-reference and full- reference image quality assessment

    Sebastian Bosse, Dominique Maniry, Klaus-Robert M¨uller, Thomas Wiegand, and Wojciech Samek. Deep neural networks for no-reference and full- reference image quality assessment. IEEE Transac- tions on Image Processing (TIP), 2017. 3

    2017

Show all 112 references
  1. [9]

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

    Jianrui Cai, Hui Zeng, Hongwei Yong, Zisheng Cao, and Lei Zhang. Toward real-world single im- age super-resolution: A new benchmark and a new model. In International Conference on Computer Vi- sion (ICCV), 2019. 16, 18

    2019

  2. [10]

    IQA-PyTorch: Py- torch toolbox for image quality assessment

    Chaofeng Chen and Jiadi Mo. IQA-PyTorch: Py- torch toolbox for image quality assessment. [On- line]. Available: https : / / github . com / chaofengc/IQA-PyTorch, 2022. 3

    2022

  3. [11]

    TOPIQ: A top-down approach from semantics to distortions for image quality assessment

    Chaofeng Chen, Jiadi Mo, Jingwen Hou, Haoning Wu, Liang Liao, Wenxiu Sun, Qiong Yan, and Weisi Lin. TOPIQ: A top-down approach from semantics to distortions for image quality assessment. IEEE Transactions on Image Processing (TIP), 2024. 3, 5, 6, 14

    2024

  4. [12]

    Human guided ground-truth generation for realistic image super-resolution

    Du Chen, Jie Liang, Xindong Zhang, Ming Liu, Hui Zeng, and Lei Zhang. Human guided ground-truth generation for realistic image super-resolution. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2023. 2, 3, 4, 6

    2023

  5. [13]

    Simple baselines for image restoration

    Liangyu Chen, Xiaojie Chu, Xiangyu Zhang, and Jian Sun. Simple baselines for image restoration. In European Conference on Computer Vision (ECCV) ,

  6. [14]

    Activating more pixels in image super-resolution transformer

    Xiangyu Chen, Xintao Wang, Jiantao Zhou, Yu Qiao, and Chao Dong. Activating more pixels in image super-resolution transformer. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR) ,

  7. [15]

    Zheng Chen, Zongwei Wu, Eduard Zamfir, Kai Zhang, Yulun Zhang, Radu Timofte, Xiaokang Yang, Hongyuan Yu, Cheng Wan, Yuxin Hong, Zhijuan Huang, Yajun Zou, Yuan Huang, Jiamin Lin, Bing- nan Han, Xianyu Guan, Yongsheng Yu, Daoan Zhang, Xuanwu Yin, Kunlong Zuo, Jinhua Hao, Kai Zha...

    2024

  8. [16]

    Directly fine-tuning diffusion models on dif- ferentiable rewards

    Kevin Clark, Paul Vicol, Kevin Swersky, and David J Fleet. Directly fine-tuning diffusion models on dif- ferentiable rewards. International Conference on Learning Representations (ICLR), 2024. 2, 3, 5

    2024

  9. [17]

    Inversion by direct iteration: An alternative to denoising diffu- sion for image restoration

    Mauricio Delbracio and Peyman Milanfar. Inversion by direct iteration: An alternative to denoising diffu- sion for image restoration. Transactions on Machine Learning Research (TMLR), 2023. 2

    2023

  10. [18]

    Image quality assessment: Unifying struc- ture and texture similarity

    Keyan Ding, Kede Ma, Shiqi Wang, and Eero P Si- moncelli. Image quality assessment: Unifying struc- ture and texture similarity. IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI) ,

  11. [19]

    Comparison of full-reference image qual- ity models for optimization of image processing sys- tems

    Keyan Ding, Kede Ma, Shiqi Wang, and Eero P Si- 9 moncelli. Comparison of full-reference image qual- ity models for optimization of image processing sys- tems. International Journal of Computer Vision (IJCV), 2021. 2, 3

    2021

  12. [20]

    Si- moncelli

    Keyan Ding, Kede Ma, Shiqi Wang, and Eero P. Si- moncelli. Image quality assessment: Unifying struc- ture and texture similarity. IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI) ,

  13. [21]

    Learning a deep convolutional network for image super-resolution

    Chao Dong, Chen Change Loy, Kaiming He, and Xi- aoou Tang. Learning a deep convolutional network for image super-resolution. In European Conference on Computer Vision (ECCV), 2014. 1, 2

    2014

  14. [22]

    Accelerating the super-resolution convolutional neu- ral network

    Chao Dong, Chen Change Loy, and Xiaoou Tang. Accelerating the super-resolution convolutional neu- ral network. In European Conference on Computer Vision (ECCV), 2016. 1, 2

    2016

  15. [23]

    DPOK: reinforcement learning for fine- tuning text-to-image diffusion models

    Ying Fan, Olivia Watkins, Yuqing Du, Hao Liu, Moonkyung Ryu, Craig Boutilier, Pieter Abbeel, Mohammad Ghavamzadeh, Kangwook Lee, and Kimin Lee. DPOK: reinforcement learning for fine- tuning text-to-image diffusion models. In Neural In- formation Processing Systems (NeurIPS), 2023. 2

    2023

  16. [24]

    Example-based super-resolution

    William T Freeman, Thouis R Jones, and Egon C Pasztor. Example-based super-resolution. IEEE Computer graphics and Applications, 2002. 1

    2002

  17. [25]

    DreamSim: Learning new dimensions of hu- man visual similarity using synthetic data

    Stephanie Fu, Netanel Tamir, Shobhita Sundaram, Lucy Chai, Richard Zhang, Tali Dekel, and Phillip Isola. DreamSim: Learning new dimensions of hu- man visual similarity using synthetic data. Neural Information Processing Systems (NeurIPS), 2023. 19

    2023

  18. [26]

    Parameter-efficient fine- tuning of state space models

    Kevin Galim, Wonjun Kang, Yuchen Zeng, Hyung Il Koo, and Kangwook Lee. Parameter-efficient fine- tuning of state space models. arXiv preprint arXiv:2410.09016, 2024. 14

    2024 arXiv

  19. [27]

    R-LPIPS: An adversarially robust perceptual similar- ity metric

    Sara Ghazanfari, Siddharth Garg, Prashanth Krish- namurthy, Farshad Khorrami, and Alexandre Araujo. R-LPIPS: An adversarially robust perceptual similar- ity metric. arXiv preprint arXiv:2307.15157, 2023. 2

    2023 arXiv

  20. [28]

    Shift-tolerant per- ceptual similarity metric

    Abhijay Ghildyal and Feng Liu. Shift-tolerant per- ceptual similarity metric. In European Conference on Computer Vision (ECCV), 2022. 6

    2022

  21. [29]

    No-reference image quality assessment via transformers, relative ranking, and self-consistency

    S Alireza Golestaneh, Saba Dadsetan, and Kris M Kitani. No-reference image quality assessment via transformers, relative ranking, and self-consistency. In Winter Conference on Applications of Computer Vision (WACV), 2022. 3

    2022

  22. [30]

    Generative adver- sarial nets

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adver- sarial nets. Neural Information Processing Systems (NeurIPS), 2014. 2

    2014

  23. [31]

    Explaining and harnessing adversarial ex- amples

    Ian J Goodfellow, Jonathon Shlens, and Christian Szegedy. Explaining and harnessing adversarial ex- amples. International Conference on Learning Rep- resentations (ICLR), 2015. 5

    2015

  24. [32]

    MambaIR: A sim- ple baseline for image restoration with state-space model

    Hang Guo, Jinmin Li, Tao Dai, Zhihao Ouyang, Xudong Ren, and Shu-Tao Xia. MambaIR: A sim- ple baseline for image restoration with state-space model. In European Conference on Computer Vision (ECCV), 2024. 2

    2024

  25. [33]

    Guardians of image quality: Benchmarking defenses against adversarial attacks on image quality metrics

    Alexander Gushchin, Khaled Abud, Georgii By- chkov, Ekaterina Shumitskaya, Anna Chistyakova, Sergey Lavrushkin, Bader Rasheed, Kirill Maly- shev, Dmitriy Vatolin, and Anastasia Antsiferova. Guardians of image quality: Benchmarking defenses against adversarial attacks on image ...

    2024

  26. [34]

    GANs trained by a two time-scale update rule con- verge to a local Nash equilibrium

    Martin Heusel, Hubert Ramsauer, Thomas Un- terthiner, Bernhard Nessler, and Sepp Hochreiter. GANs trained by a two time-scale update rule con- verge to a local Nash equilibrium. Neural Informa- tion Processing Systems (NeurIPS), 2017. 3

    2017

  27. [35]

    KonIQ-10k: An ecologically valid database for deep learning of blind image quality assessment

    Vlad Hosu, Hanhe Lin, Tamas Sziranyi, and Dietmar Saupe. KonIQ-10k: An ecologically valid database for deep learning of blind image quality assessment. IEEE Transactions on Image Processing (TIP), 2020. 3, 14

    2020

  28. [36]

    LoRA: Low-rank adaptation of large language models

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. LoRA: Low-rank adaptation of large language models. International Conference on Learning Representations (ICLR), 2022. 5, 14

    2022

  29. [37]

    Real-world super- resolution via kernel estimation and noise injection

    Xiaozhong Ji, Yun Cao, Ying Tai, Chengjie Wang, Jilin Li, and Feiyue Huang. Real-world super- resolution via kernel estimation and noise injection. In IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2020. 4

    2020

  30. [38]

    Tackling the ill-posedness of super- resolution through adaptive target generation

    Younghyun Jo, Seoung Wug Oh, Peter Vajda, and Seon Joo Kim. Tackling the ill-posedness of super- resolution through adaptive target generation. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2021. 2

    2021

  31. [39]

    Per- ceptual losses for real-time style transfer and super- resolution

    Justin Johnson, Alexandre Alahi, and Li Fei-Fei. Per- ceptual losses for real-time style transfer and super- resolution. In European Conference on Computer Vi- sion (ECCV), 2016. 2

    2016

  32. [40]

    Denoising diffusion restoration models

    Bahjat Kawar, Michael Elad, Stefano Ermon, and Ji- aming Song. Denoising diffusion restoration models. Neural Information Processing Systems (NeurIPS) ,

  33. [41]

    MUSIQ: Multi-scale image qual- ity transformer

    Junjie Ke, Qifei Wang, Yilin Wang, Peyman Milan- far, and Feng Yang. MUSIQ: Multi-scale image qual- ity transformer. InInternational Conference on Com- puter Vision (ICCV), 2021. 2, 3, 5, 6, 14 10

    2021

  34. [42]

    E-LPIPS: robust perceptual image similarity via random transformation ensembles

    Markus Kettunen, Erik H ¨ark¨onen, and Jaakko Lehti- nen. E-LPIPS: robust perceptual image similarity via random transformation ensembles. arXiv preprint arXiv:1906.03973, 2019. 2

    1906 arXiv

  35. [43]

    Neural side-by-side: Predicting human preferences for no- reference super-resolution evaluation

    Valentin Khrulkov and Artem Babenko. Neural side-by-side: Predicting human preferences for no- reference super-resolution evaluation. In IEEE Con- ference on Computer Vision and Pattern Recognition (CVPR), 2021. 3, 14

    2021

  36. [44]

    Deeply-recursive convolutional network for image super-resolution

    Jiwon Kim, Jung Kwon Lee, and Kyoung Mu Lee. Deeply-recursive convolutional network for image super-resolution. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016. 2

    2016

  37. [45]

    Adversarial machine learning at scale

    Alexey Kurakin, Ian Goodfellow, and Samy Ben- gio. Adversarial machine learning at scale. In- ternational Conference on Learning Representations (ICLR), 2017. 5

    2017

  38. [46]

    DataInf: Efficiently estimating data influence in LoRA-tuned LLMs and diffusion models

    Yongchan Kwon, Eric Wu, Kevin Wu, and James Zou. DataInf: Efficiently estimating data influence in LoRA-tuned LLMs and diffusion models. In- ternational Conference on Learning Representations (ICLR), 2024. 14

    2024

  39. [47]

    Deep Laplacian pyramid net- works for fast and accurate super-resolution

    Wei-Sheng Lai, Jia-Bin Huang, Narendra Ahuja, and Ming-Hsuan Yang. Deep Laplacian pyramid net- works for fast and accurate super-resolution. InIEEE Conference on Computer Vision and Pattern Recog- nition (CVPR), 2017. 2

    2017

  40. [48]

    Photo-realistic single image super- resolution using a generative adversarial network

    Christian Ledig, Lucas Theis, Ferenc Husz ´ar, Jose Caballero, Andrew Cunningham, Alejandro Acosta, Andrew Aitken, Alykhan Tejani, Johannes Totz, Ze- han Wang, et al. Photo-realistic single image super- resolution using a generative adversarial network. In IEEE Conference on C...

    2017

  41. [49]

    Learning generative structure prior for blind text image super-resolution

    Xiaoming Li, Wangmeng Zuo, and Chen Change Loy. Learning generative structure prior for blind text image super-resolution. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR) ,

  42. [50]

    SwinIR: Image restoration using swin transformer

    Jingyun Liang, Jiezhang Cao, Guolei Sun, Kai Zhang, Luc Van Gool, and Radu Timofte. SwinIR: Image restoration using swin transformer. InInterna- tional Conference on Computer Vision (ICCV), 2021. 2, 6, 14

    2021

  43. [51]

    Details or ar- tifacts: A locally discriminative learning approach to realistic image super-resolution

    Jie Liang, Hui Zeng, and Lei Zhang. Details or ar- tifacts: A locally discriminative learning approach to realistic image super-resolution. In IEEE Con- ference on Computer Vision and Pattern Recognition (CVPR), 2022. 2

    2022

  44. [52]

    Rich hu- man feedback for text-to-image generation

    Youwei Liang, Junfeng He, Gang Li, Peizhao Li, Ar- seniy Klimovskiy, Nicholas Carolan, Jiao Sun, Jordi Pont-Tuset, Sarah Young, Feng Yang, et al. Rich hu- man feedback for text-to-image generation. In IEEE Conference on Computer Vision and Pattern Recog- nition (CVPR), 2024. 3

    2024

  45. [53]

    Tracking meets LoRA: Faster training, larger model, stronger perfor- mance

    Liting Lin, Heng Fan, Zhipeng Zhang, Yaowei Wang, Yong Xu, and Haibin Ling. Tracking meets LoRA: Faster training, larger model, stronger perfor- mance. In European Conference on Computer Vision (ECCV), 2024. 14

    2024

  46. [54]

    VMamba: Visual state space model

    Yue Liu, Yunjie Tian, Yuzhong Zhao, Hongtian Yu, Lingxi Xie, Yaowei Wang, Qixiang Ye, Jianbin Jiao, and Yunfan Liu. VMamba: Visual state space model. In Neural Information Processing Systems (NeurIPS), 2024. 2

    2024

  47. [55]

    Defense against adversarial at- tacks on no-reference image quality models with gra- dient norm regularization

    Yujia Liu, Chenxi Yang, Dingquan Li, Jianhao Ding, and Tingting Jiang. Defense against adversarial at- tacks on no-reference image quality models with gra- dient norm regularization. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR) ,

  48. [56]

    SRFlow: Learning the super- resolution space with normalizing flow

    Andreas Lugmayr, Martin Danelljan, Luc Van Gool, and Radu Timofte. SRFlow: Learning the super- resolution space with normalizing flow. In European Conference on Computer Vision (ECCV), 2020. 2

    2020

  49. [57]

    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 (CVIU), 2017. 3

    2017

  50. [58]

    Structure-preserving super resolution with gradient guidance

    Cheng Ma, Yongming Rao, Yean Cheng, Ce Chen, Jiwen Lu, and Jie Zhou. Structure-preserving super resolution with gradient guidance. In IEEE Confer- ence on Computer Vision and Pattern Recognition (CVPR), 2020. 2

    2020

  51. [59]

    No-reference image quality assessment in the spatial domain

    Anish Mittal, Anush Krishna Moorthy, and Alan Conrad Bovik. No-reference image quality assessment in the spatial domain. IEEE Transactions on Image Processing (TIP), 2012. 3, 14

    2012

  52. [60]

    Under- standing SSIM

    Jim Nilsson and Tomas Akenine-M ¨oller. Under- standing SSIM. arXiv preprint arXiv:2006.13846 ,

    2006 arXiv

  53. [61]

    Exploiting deep generative prior for versatile image restoration and manipulation

    Xingang Pan, Xiaohang Zhan, Bo Dai, Dahua Lin, Chen Change Loy, and Ping Luo. Exploiting deep generative prior for versatile image restoration and manipulation. IEEE Transactions on Pattern Analy- sis and Machine Intelligence (PAMI), 2021. 2

    2021

  54. [62]

    Content-aware local GAN for photo-realistic super- resolution

    JoonKyu Park, Sanghyun Son, and Kyoung Mu Lee. Content-aware local GAN for photo-realistic super- resolution. In International Conference on Computer Vision (ICCV), 2023. 2

    2023

  55. [63]

    Perception-oriented single image super-resolution using optimal objective estimation

    Seung Ho Park, Young Su Moon, and Nam Ik Cho. Perception-oriented single image super-resolution using optimal objective estimation. In IEEE Con- ference on Computer Vision and Pattern Recognition (CVPR), 2023. 2 11

    2023

  56. [64]

    Understanding mid-level rep- resentations in visual processing

    Jonathan W Peirce. Understanding mid-level rep- resentations in visual processing. Journal of Vision (JOV), 2015. 19

    2015

  57. [65]

    Perception viewed as an inverse problem

    Zygmunt Pizlo. Perception viewed as an inverse problem. Vision research, 41(24):3145–3161, 2001. 1

    2001

  58. [66]

    Color image database TID2013: Peculiarities and preliminary results

    Nikolay Ponomarenko, Oleg Ieremeiev, Vladimir Lukin, Karen Egiazarian, Lina Jin, Jaakko Astola, Benoit V ozel, Kacem Chehdi, Marco Carli, Feder- ica Battisti, et al. Color image database TID2013: Peculiarities and preliminary results. In European workshop on visual information...

    2013

  59. [67]

    Aligning text-to-image diffusion models with reward backpropagation

    Mihir Prabhudesai, Anirudh Goyal, Deepak Pathak, and Katerina Fragkiadaki. Aligning text-to-image diffusion models with reward backpropagation. arXiv preprint arXiv:2310.03739, 2023. 2

    2023 arXiv

  60. [68]

    A DCT statistics-based blind image quality index

    Michele A Saad, Alan C Bovik, and Christophe Charrier. A DCT statistics-based blind image quality index. IEEE Signal Processing Letters, 2010. 2, 3

    2010

  61. [69]

    Image super-resolution via iterative refine- ment

    Chitwan Saharia, Jonathan Ho, William Chan, Tim Salimans, David J Fleet, and Mohammad Norouzi. Image super-resolution via iterative refine- ment. IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), 2022. 2

    2022

  62. [70]

    Generative adver- sarial networks (GANs) challenges, solutions, and future directions

    Divya Saxena and Jiannong Cao. Generative adver- sarial networks (GANs) challenges, solutions, and future directions. ACM Computing Surveys (CSUR),

  63. [71]

    Real-time single im- age and video super-resolution using an efficient sub- pixel convolutional neural network

    Wenzhe Shi, Jose Caballero, Ferenc Husz ´ar, Jo- hannes Totz, Andrew P Aitken, Rob Bishop, Daniel Rueckert, and Zehan Wang. Real-time single im- age and video super-resolution using an efficient sub- pixel convolutional neural network. In IEEE Con- ference on Computer Vision a...

    2016

  64. [72]

    Blindly assess image quality in the wild guided by a self-adaptive hyper network

    Shaolin Su, Qingsen Yan, Yu Zhu, Cheng Zhang, Xin Ge, Jinqiu Sun, and Yanning Zhang. Blindly assess image quality in the wild guided by a self-adaptive hyper network. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2020. 3

    2020

  65. [73]

    Intriguing properties of neural networks

    Christian Szegedy, Wojciech Zaremba, Ilya Sutskever, Joan Bruna, Dumitru Erhan, Ian Goodfel- low, and Rob Fergus. Intriguing properties of neural networks. In International Conference on Learning Representations (ICLR), 2014. 5

    2014

  66. [74]

    NIMA: Neu- ral image assessment

    Hossein Talebi and Peyman Milanfar. NIMA: Neu- ral image assessment. IEEE Transactions on Image Processing (TIP), 2018. 2, 3, 5, 6

    2018

  67. [75]

    Learning a blind measure of perceptual image qual- ity

    Huixuan Tang, Neel Joshi, and Ashish Kapoor. Learning a blind measure of perceptual image qual- ity. InIEEE Conference on Computer Vision and Pat- tern Recognition (CVPR), 2011. 2, 3

    2011

  68. [76]

    Anchored neighborhood regression for fast example- based super-resolution

    Radu Timofte, Vincent De Smet, and Luc Van Gool. Anchored neighborhood regression for fast example- based super-resolution. In International Conference on Computer Vision (ICCV), 2013. 2

    2013

  69. [77]

    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 Proceedings of the Asian Conference on Computer Vision (ACCV), 2015. 2

    2015

  70. [78]

    Wasserstein auto-encoders

    Ilya Tolstikhin, Olivier Bousquet, Sylvain Gelly, and Bernhard Schoelkopf. Wasserstein auto-encoders. arXiv preprint arXiv:1711.01558v3, 2018. 19

    2018 arXiv

  71. [79]

    Deep cyclic generative adversarial residual convo- lutional networks for real image super-resolution

    Rao Muhammad Umer and Christian Micheloni. Deep cyclic generative adversarial residual convo- lutional networks for real image super-resolution. In European Conference on Computer Vision Work- shops (ECCVW), 2020. 2

    2020

  72. [80]

    Methodol- ogy for the subjective assessment of the quality of television pictures

    International Telecommunication Union. Methodol- ogy for the subjective assessment of the quality of television pictures. Recommendation ITU-R BT.500- 13, 2012. Geneva, Switzerland. 18

    2012

  73. [81]

    Blind image quality evaluation using perception based fea- tures

    Narasimhan Venkatanath, D Praneeth, Maruthi Chandrasekhar Bh, Sumohana S Chan- nappayya, and Swarup S Medasani. Blind image quality evaluation using perception based fea- tures. In 2015 Twenty first national conference on communications (NCC). IEEE, 2015. 3

    2015

  74. [82]

    Exploring CLIP for assessing the look and feel of images

    Jianyi Wang, Kelvin CK Chan, and Chen Change Loy. Exploring CLIP for assessing the look and feel of images. In Proceedings of the National Confer- ence on Artificial Intelligence (AAAI), 2023. 3

    2023

  75. [83]

    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 (IJCV) ,

  76. [84]

    ESRGAN: Enhanced super-resolution generative ad- versarial networks

    Xintao Wang, Ke Yu, Shixiang Wu, Jinjin Gu, Yihao Liu, Chao Dong, Yu Qiao, and Chen Change Loy. ESRGAN: Enhanced super-resolution generative ad- versarial networks. InEuropean Conference on Com- puter Vision Workshops (ECCVW), 2018. 2, 6, 14

    2018

  77. [85]

    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. InInterna- tional Conference on Computer Vision (ICCV), 2021. 4, 6, 14

    2021

  78. [86]

    Image quality assessment: from error visibility to structural similarity

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Simoncelli. Image quality assessment: from error visibility to structural similarity. IEEE Trans- actions on Image Processing (TIP), 2004. 2, 6

    2004

  79. [87]

    Exploring video quality assessment on user generated contents from aesthetic 12 and technical perspectives

    Haoning Wu, Erli Zhang, Liang Liao, Chaofeng Chen, Jingwen Hou, Annan Wang, Wenxiu Sun, Qiong Yan, and Weisi Lin. Exploring video quality assessment on user generated contents from aesthetic 12 and technical perspectives. In International Confer- ence on Computer Vision (ICCV)...

    2023

  80. [88]

    Q- Align: Teaching LMMs for visual scoring via dis- crete text-defined levels

    Haoning Wu, Zicheng Zhang, Weixia Zhang, Chaofeng Chen, Liang Liao, Chunyi Li, Yixuan Gao, Annan Wang, Erli Zhang, Wenxiu Sun, Qiong Yan, Xiongkuo Min, Guangtao Zhai, and Weisi Lin. Q- Align: Teaching LMMs for visual scoring via dis- crete text-defined levels. InInternational ...

    2024

  81. [89]

    One-step effective diffusion network for real- world image super-resolution

    Rongyuan Wu, Lingchen Sun, Zhiyuan Ma, and Lei Zhang. One-step effective diffusion network for real- world image super-resolution. Neural Information Processing Systems (NeurIPS), 2024. 2, 18

    2024

  82. [90]

    SeeSR: Towards semantics-aware real-world image super-resolution

    Rongyuan Wu, Tao Yang, Lingchen Sun, Zhengqiang Zhang, Shuai Li, and Lei Zhang. SeeSR: Towards semantics-aware real-world image super-resolution. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2024. 18

    2024

  83. [91]

    DiffIR: Efficient diffusion model for im- age restoration

    Bin Xia, Yulun Zhang, Shiyin Wang, Yitong Wang, Xinglong Wu, Yapeng Tian, Wenming Yang, and Luc Van Gool. DiffIR: Efficient diffusion model for im- age restoration. In International Conference on Com- puter Vision (ICCV), 2023. 2

    2023

  84. [92]

    ImageReward: Learning and evaluating human pref- erences for text-to-image generation

    Jiazheng Xu, Xiao Liu, Yuchen Wu, Yuxuan Tong, Qinkai Li, Ming Ding, Jie Tang, and Yuxiao Dong. ImageReward: Learning and evaluating human pref- erences for text-to-image generation. Neural Infor- mation Processing Systems (NeurIPS), 2024. 2, 3

    2024

  85. [93]

    Coupled dictionary train- ing for image super-resolution

    Jianchao Yang, Zhaowen Wang, Zhe Lin, Scott Co- hen, and Thomas Huang. Coupled dictionary train- ing for image super-resolution. IEEE Transactions on Image Processing (TIP), 2012. 2

    2012

  86. [94]

    MANIQA: Multi-dimension attention network for no-reference image quality assessment

    Sidi Yang, Tianhe Wu, Shuwei Shi, Shanshan Lao, Yuan Gong, Mingdeng Cao, Jiahao Wang, and Yujiu Yang. MANIQA: Multi-dimension attention network for no-reference image quality assessment. In IEEE Conference on Computer Vision and Pattern Recog- nition (CVPR), 2022. 3

    2022

  87. [95]

    Pixel-aware stable diffusion for realistic image super-resolution and personalized stylization

    Tao Yang, Rongyuan Wu, Peiran Ren, Xuansong Xie, and Lei Zhang. Pixel-aware stable diffusion for realistic image super-resolution and personalized stylization. In European Conference on Computer Vision (ECCV), 2024. 2, 18

    2024

  88. [96]

    Local implicit normalizing flow for arbitrary-scale image super- resolution

    Jie-En Yao, Li-Yuan Tsao, Yi-Chen Lo, Roy Tseng, Chia-Che Chang, and Chun-Yi Lee. Local implicit normalizing flow for arbitrary-scale image super- resolution. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2023. 2

    2023

  89. [97]

    From patches to pictures (PaQ-2-PiQ): Mapping the per- ceptual space of picture quality

    Zhenqiang Ying, Haoran Niu, Praful Gupta, Dhruv Mahajan, Deepti Ghadiyaram, and Alan Bovik. From patches to pictures (PaQ-2-PiQ): Mapping the per- ceptual space of picture quality. In IEEE Confer- ence on Computer Vision and Pattern Recognition (CVPR), 2020. 3, 5, 14

    2020

  90. [98]

    A feature- enriched completely blind image quality evaluator

    Lin Zhang, Lei Zhang, and Alan C Bovik. A feature- enriched completely blind image quality evaluator. IEEE Transactions on Image Processing (TIP), 2015. 3, 14

    2015

  91. [99]

    The unreasonable ef- fectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shechtman, and Oliver Wang. The unreasonable ef- fectiveness of deep features as a perceptual metric. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018. 2

    2018

  92. [100]

    Blind image quality assessment using a deep bilinear convolutional neural network

    Weixia Zhang, Kede Ma, Jia Yan, Dexiang Deng, and Zhou Wang. Blind image quality assessment using a deep bilinear convolutional neural network. IEEE Transactions on Circuits and Systems for Video Tech- nology, 2020. 3

    2020

  93. [101]

    Blind image quality assess- ment via vision-language correspondence: A mul- titask learning perspective

    Weixia Zhang, Guangtao Zhai, Ying Wei, Xiaokang Yang, and Kede Ma. Blind image quality assess- ment via vision-language correspondence: A mul- titask learning perspective. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR) ,

  94. [102]

    Image super-resolution us- ing very deep residual channel attention networks

    Yulun Zhang, Kunpeng Li, Kai Li, Lichen Wang, Bi- neng Zhong, and Yun Fu. Image super-resolution us- ing very deep residual channel attention networks. In European Conference on Computer Vision (ECCV) ,

  95. [103]

    Diffusion-based blind text image super-resolution

    Yuzhe Zhang, Jiawei Zhang, Hao Li, Zhouxia Wang, Luwei Hou, Dongqing Zou, and Liheng Bian. Diffusion-based blind text image super-resolution. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2024. 19

    2024

  96. [104]

    Learning conditional knowledge distillation for degraded-reference image quality as- sessment

    Heliang Zheng, Huan Yang, Jianlong Fu, Zheng-Jun Zha, and Jiebo Luo. Learning conditional knowledge distillation for degraded-reference image quality as- sessment. In International Conference on Computer Vision (ICCV), 2021. 3 13 Augmenting Perceptual Super-Resolution via Imag...

    2021

  97. [105]

    Here, in Table 6, we present accuracy of 20 NR-IQA met- rics and their variants (42 in total) on the same subset

    Complete Analysis of NR-IQA metrics In §3.1 of the main paper, we present accuracy of only top 7 NR-IQA metrics on the subset of SBS180K [43] train set. Here, in Table 6, we present accuracy of 20 NR-IQA met- rics and their variants (42 in total) on the same subset. We make tw...

  98. [106]

    FR Low-Lev Dist

    Methodological Details 8.1. Sampling Details The altered sampling (§4.2) is trained identically to the stan- dard HGGT version, just replacing the uniform nature of the GT sampling. The only additional parameter is the temper- ature,τ, which we set to 10 for both SMA and SMP. ...

  99. [107]

    FTNNR,IG×2

    Detailed Results on Ablations and Variations In this section, we consider additional FT variations: (i) us- ing a GAN discriminator instead of an NR-IQA model (us- ing two different loss weights) and (ii) replacing Q (set as MUSIQ) with a different NR-IQA model (PaQ-2-PiQ). Th...

  100. [108]

    grid-like

    Additional Qualitative Examples 10.1. Additional Comparative Samples Additional comparisons are shown in Fig. 6 (as in Fig. 4). Our method (AMO or AMO+FT) is universally sharper and more detailed than UPos (e.g., see the hair in row three). Further, it can occasionally remove ...

  101. [109]

    Additional Results on RealSR We provide results on the RealSRv3 [9] dataset in Table 8. Similar to the HGGT test dataset, we find thatour method is superior in terms of every NR-IQA metric, at the expense of the exact pixel-level details measured by PSNR and SSIM (following th...

  102. [110]

    AMO” and “AMO + FT

    Comparative Evaluation via User Study Similar to the HGGT user study, we invite 12 volunteers to evaluate their preference between SwinIR-AMO+FT and SwinIR-UPos, using the HGGT Test-100 dataset. Each volunteer evaluates 25 image pairs (25% of the dataset), with each image in T...

  103. [111]

    groupings

    Remark on Evaluation Metric Types and Nomenclature The perception-distortion tradeoff [5] necessitates a com- plex suite of evaluation metrics that consider different as- pects of the SR outputs, including pixel-level fidelity to a GT image and standalone image quality. Some w...

  104. [112]

    First, higher IQA model score does not guarantee improved human perceptual quality nor does it strictly en- sure our outputs are artifact-free

    Limitations While our IQA-based method is able to sharpen SR out- puts, as well as hallucinate aesthetically pleasing details in most cases, there are still several shortcomings to our ap- proach. First, higher IQA model score does not guarantee improved human perceptual quali...

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.