Pith. sign in

REVIEW 6 major objections 5 minor 54 references

Scene Perceived Image Perceptual Score (SPIPS): combining global and local perception for image quality assessment

T0 review · 6 major / 5 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 SPIPS, a hybrid IQA metric that separates semantic and perceptual deep features and adds traditional scores, reports the highest agreement with human judgment among compared methods on every BAPPS 2AFC category.

desk verdict A sensible hybrid IQA metric whose main experimental claim is currently undercut by an undefined 2AFC correlation protocol and a contradictory 'data-free' claim; worth a serious round of revision, not a rejection. read the letter →

arxiv 2504.17234 v1 pith:NUBNQCKS submitted 2025-04-24 cs.CV

Zhiqiang Lao , Heather Yu This is my paper · ORCID
classification cs.CV
keywords imagequalityassessmentfull-referenceIQAperceptualmetricdeepfeaturessemanticBAPPSdatasetLPIPSSSIM
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

SPIPS is a full-reference image quality metric that fuses three evidence streams: classic pixel and structural scores (PSNR, SSIM, MS-SSIM), low-level perceptual deep features, and high-level semantic deep features from a pretrained CNN. The paper's thesis is that distinguishing semantic content from perceptual detail, and adding traditional metrics, yields judgments closer to human perception than any single-stream metric. On the BAPPS two-alternative forced-choice benchmark, SPIPS reports the highest correlation with human ratings in every distortion category among the tested methods, and it generally leads on the just-noticeable-difference subset as well. If these results hold, SPIPS offers a straightforward recipe for scoring AI-generated or heavily processed images in line with what people actually see.

What carries the argument

The load-bearing object is the layer partition in Eqs. (7)-(8): for a backbone with $L$ feature extraction layers, the first $L-2$ layers are the 'perceptual' group and the last two are the 'semantic' group, and the squared difference between reference and evaluated features at each layer produces a quality map that enters one of those two streams. The second mechanism is the fusion rule in Eq. (12), where the mean activations of three independently convolved streams—traditional, perceptual, and semantic—are combined with learned weights summing to one. The partition operationalizes the paper's top-down account of human vision, in which overall structure is grasped before fine details, and the weighted fusion lets the network balance global content against local artifacts when producing a single score.

What would settle it

Train SPIPS on BAPPS with the perceptual/semantic boundary moved to $L-1$, $L-3$, or a learned per-layer weighting, and compare SRCC on the 2AFC test set; if any alternative split matches or exceeds the reported numbers, the specific claim that the last two layers are uniquely semantic is falsified.

Watch

Extended reading notes

Core claim

SPIPS's central claim is that image quality is best measured as a blend of separable streams rather than one global difference. From a pretrained CNN, the squared element-wise differences between the evaluated and reference feature maps are partitioned by layer: the first $L-2$ layers produce 'perceptual' quality maps and the last two layers produce 'semantic' quality maps. In parallel, per-pixel maps from PSNR, SSIM, and MS-SSIM are inverted so that lower values mean better quality. Each of the three map groups passes through its own small convolutional feature extractor with ReLU, and the mean of each extracted stream is combined in a weighted sum, with weights $\lambda_1,\lambda_2,\lambda_3$ and the small convolutions trained on the BAPPS dataset following the LPIPS training protocol. The paper reports that the resulting score achieves the highest PLCC, SRCC, and KRCC across all six BAPPS 2AFC categories when compared with PSNR, SSIM, VIF, DISTS, and LPIPS, and the best SRCC and KRCC on the JND split.

Load-bearing premise

The load-bearing premise is that the fixed division of network layers into the first $L-2$ (perceptual) and last two (semantic) mirrors the human visual system's global-before-local judgment; the paper offers no ablation or evidence for alternatives to this split.

Editorial extensions

If this is right

  • On BAPPS 2AFC, SPIPS reports the highest PLCC, SRCC, and KRCC in all six distortion categories, beating PSNR, SSIM, VIF, DISTS, and LPIPS.
  • On the BAPPS JND subset, SPIPS achieves the best SRCC and KRCC on the CNN and Trad splits, while its PLCC on the CNN split is slightly below LPIPS and DISTS.
  • Ablations that remove either the semantic stream or the traditional metrics reduce correlation on every reported metric, indicating that both streams contribute to the final score.
  • Because the backbone stays pretrained and only the fusion weights and small per-stream convolutions are learned, adapting SPIPS to a new distortion family requires comparatively little training.

Reading between the lines

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

  • A learnable layer split could reveal whether the semantic/perceptual boundary actually sits at different depths for different distortion types, and might improve the scores further; the fixed $L-2$ split in the paper is only one point in that design space.
  • The largest reported gains over the next best metric appear on colorization and traditional distortions, which suggests the traditional-metric stream carries the most weight where deep features are weakest; a per-category ablation of that stream would test this directly.
  • The same three-stream disentanglement could transfer to no-reference quality assessment or video quality, and the paper's stated use of vision transformers would let the semantic stream operate on token relationships rather than layer groups.
  • The per-category reporting leaves the practitioner to choose how to aggregate; a natural next step is a single pooled correlation over all BAPPS splits, which the paper does not provide.
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

6 major / 5 minor

Summary. The paper proposes SPIPS, a full-reference image quality assessment (IQA) metric that fuses three streams: per-pixel traditional IQA maps (PSNR, SSIM, MS-SSIM), low-level perceptual deep features, and high-level semantic deep features extracted from a pretrained CNN. The deep features are split by layer index into perceptual and semantic groups, each passed through its own convolution block, and the final score is a weighted combination of the three streams. The authors train the model on the BAPPS dataset and report PLCC, SRCC, and KRCC on the BAPPS 2AFC and JND subsets, claiming that SPIPS consistently achieves the highest correlation with human judgments among compared methods (PSNR, SSIM, VIF, DISTS, LPIPS). The paper presents qualitative examples, two ablation studies, and a conclusion outlining future work.

Significance. If the reported results are reproducible and the evaluation protocol is well defined, the central idea of explicitly separating semantic from perceptual deep features and combining them with traditional metrics is a plausible and potentially useful contribution to full-reference IQA, especially for DNN-generated and DNN-processed images. The qualitative examples in Figures 3 and 4 are illustrative, and the ablation study is the right kind of evidence for the fusion design. However, the manuscript currently contains a direct contradiction between the claimed 'data-free' / 'no training' status of the method and the explicit description of training on BAPPS, and it leaves the evaluation protocol for 2AFC data unspecified. These issues prevent verification of the central claim and must be resolved before the contribution can be assessed.

major comments (6)
  1. [Section IV.A, Section IV.B, Tables V-VII] The evaluation protocol for the 2AFC subset is not defined. BAPPS 2AFC provides binary human preferences (which of two images is closer to the reference), not scalar quality labels. PLCC, SRCC, and KRCC require scalar ground-truth values, and the paper never states how the binary preferences are converted into the ground-truth vectors used in Tables V-VII, nor how the model's scores for image0 and image1 are aggregated into a per-trial or per-image prediction. Without this description, 'our model consistently achieves the highest correlation across all categories' is not a well-defined, reproducible claim. The authors should specify the exact aggregation and correlation computation, and note whether the values are comparable to standard IQA PLCC/SRCC numbers.
  2. [Section II.A, Section II.B, Section IV.A, Figure 2 caption] The claim that the method is 'data-free' and 'requires no training at all' is contradicted by Section IV.A ('BAPPS ... is used to train and validate SPIPS model') and by the Figure 2 caption ('The SPIPS model parameters are iteratively optimized via backpropagation'). This is not a wording issue: it changes the interpretation of the experimental comparison, since SPIPS is a trained model while several baselines are fixed metrics. The authors must remove the data-free claim or clearly specify what is trained and what is fixed, including the training set split.
  3. [Section III.A, Equations (1)-(3)] Per-pixel PSNR and per-pixel MS-SSIM maps are not standard definitions and are not defined in the paper. PSNR is normally a global scalar computed from MSE, and MS-SSIM is a scalar or, in some multi-scale formulations, a map of per-pixel local scores, but the paper does not give the formula used, and Eq. (3) states Q_msssim has C channels (the number of scales) while each Q_p value is claimed to be in R^{3xHxW}. The normalization N(·) is also unspecified. Without these definitions, the traditional IQA module cannot be reproduced, and the reported maps in Figures 3 and 4 are not interpretable.
  4. [Section III.B, Equations (7)-(8)] The split of CNN layers into 'perceptual' (first L-2 layers) and 'semantic' (last two layers) is load-bearing for the entire fusion design, but the paper offers no justification or ablation for this choice. It is stated as an example for AlexNet with no evidence that the split is optimal or even robust across backbones. An ablation over alternative splits (e.g., different cut points, or learned layer weights) is needed to support the claim that this specific distinction is what drives the reported gains.
  5. [Section IV.C, Tables II-IV] The ablation study reports only point estimates (e.g., SPIPS-abla1 vs. SPIPS on CNN: 0.78 vs. 0.81 in PLCC) with no error bars, number of trials, statistical significance tests, or details about training seeds. Many differences are small (0.01-0.05) and could be within run-to-run variation. The authors should report multiple runs or at least a significance analysis to support the statement that 'both ablation models performed worse than the full model across all metrics.'
  6. [Section IV.D, Table VIII] The claim of 'consistently highest correlation' is not consistent with the JND results on the CNN category: SPIPS (PLCC 0.60) is lower than both LPIPS (0.63) and DISTS (0.63). The text acknowledges this in passing, but the abstract and Section IV.D's opening claim of consistent superiority should be reconciled with these numbers, and the conditions under which the method is superior should be stated precisely.
minor comments (5)
  1. [References] Reference [6] is truncated: 'Transactions on Machine Le' should be completed with the journal name and volume/page details.
  2. [Section III.B, text after Eq. (5)] The text says 'height and weight' where it should say 'height and width'; also the feature map dimensions for different layers are never stated, so it is unclear how the error maps of different resolutions are aligned before convolution.
  3. [Section III.D, Eq. (12)] The constraint lambda_1 + lambda_2 + lambda_3 = 1 is stated, but no non-negativity constraint is imposed; the meaning of negative weights in the final score should be discussed or the weights should be constrained.
  4. [Figures 3 and 4] The captions do not explicitly state that lower SPIPS scores indicate better quality, which is essential for interpreting the numerical values in the figures; this should be stated in each caption.
  5. [Section IV.A] The paper says BAPPS 'is used to train and validate SPIPS model' but never specifies the train/validation/test split, whether the 'test split' mentioned in Section IV.B refers to the official BAPPS test split, and whether any baselines were retrained or used with their published settings.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: SPIPS is trained on BAPPS training labels and evaluated on a held-out test split; the claimed results are empirical, not forced by construction.

full rationale

The derivation chain in SPIPS is empirical rather than circular. The model is explicitly defined by Eqs. (1)-(12): traditional IQA maps (PSNR, SSIM, MS-SSIM), squared differences of deep features from pretrained networks, per-group convolutions, and a weighted sum. The only learned parameters (the convolutions and λ weights) are optimized using BAPPS human labels on a training split, as stated in the Figure 2 caption ('The SPIPS model parameters are iteratively optimized via backpropagation'), and then evaluated on a held-out BAPPS test split in Tables II-VIII. Testing on held-out data after supervised training is standard evaluation, not a prediction that reduces to its inputs. The layer split in Eqs. (7)-(8) is an unvalidated design heuristic, but it is an assumption about network layer semantics, not a self-referential definition of the quality score; the appeal to human visual hierarchy is motivation, not a derivation. I find no load-bearing self-citation: the only overlapping reference, [35], is not cited in the body text and does not support any central premise. The paper does contain an internal contradiction—Section II.B claims 'our method requires no training at all' while Section IV.A says BAPPS 'is used to train and validate SPIPS model'—and the 2AFC PLCC/SRCC/KRCC protocol is under-specified for binary-preference data. These are correctness and verifiability concerns, not circularity. No identified step makes a claimed result equivalent to its own inputs by construction.

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

The central claim depends on a handful of learned weights and convolutions, a heuristic layer split, and the validity of BAPPS as ground truth. No new physical or mathematical entities are introduced.

free parameters (3)
  • lambda_1, lambda_2, lambda_3 = not reported
    Weighting terms for tradition, percept, and semantic streams in Eq. 12. Learned during training on BAPPS.
  • Convolution weights C_tradition, C_percept, C_semantic = not reported
    Convolution operations in Eqs. 9-11 extract quality features from each map group. Learned via backpropagation.
  • Layer split threshold (L-2) = L-2
    Heuristic choice to separate perceptual (layers 1..L-2) from semantic (layers L-1, L) features. Not learned but chosen by hand without ablation.
assumptions (4)
  • domain assumption Pretrained CNN features provide perceptually relevant representations for image quality.
    Borrowed from LPIPS literature [5]; the paper relies on this to justify using AlexNet/VGG features without task-specific pretraining.
  • domain assumption BAPPS human annotations are valid ground truth for perceptual quality.
    The model is trained and evaluated on BAPPS; the paper treats its labels as the objective measure of human perception.
  • ad hoc to paper The split of CNN layers into perceptual and semantic groups mirrors human global/local processing.
    Introduced in Section III-B, Eqs. 7-8, with no empirical or theoretical justification for the specific L-2 cutoff.
  • domain assumption Per-pixel PSNR, SSIM, and MS-SSIM maps can be defined and are meaningful.
    Eqs. 1-3 assume these metrics can produce spatial maps, but standard PSNR and MS-SSIM are global scalars; no construction is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Scene Perceived Image Perceptual Score (SPIPS): combining global and local perception for image quality assessment." pith.science (2026). https://pith.science/paper/NUBNQCKS

@misc{pith2026250417234,
  author       = {Pith},
  title        = {Pith review of: Scene Perceived Image Perceptual Score (SPIPS): combining global and local perception for image quality assessment},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NUBNQCKS}},
  note         = {Machine review of arXiv:2504.17234}
}
read the original abstract

The rapid advancement of artificial intelligence and widespread use of smartphones have resulted in an exponential growth of image data, both real (camera-captured) and virtual (AI-generated). This surge underscores the critical need for robust image quality assessment (IQA) methods that accurately reflect human visual perception. Traditional IQA techniques primarily rely on spatial features - such as signal-to-noise ratio, local structural distortions, and texture inconsistencies - to identify artifacts. While effective for unprocessed or conventionally altered images, these methods fall short in the context of modern image post-processing powered by deep neural networks (DNNs). The rise of DNN-based models for image generation, enhancement, and restoration has significantly improved visual quality, yet made accurate assessment increasingly complex. To address this, we propose a novel IQA approach that bridges the gap between deep learning methods and human perception. Our model disentangles deep features into high-level semantic information and low-level perceptual details, treating each stream separately. These features are then combined with conventional IQA metrics to provide a more comprehensive evaluation framework. This hybrid design enables the model to assess both global context and intricate image details, better reflecting the human visual process, which first interprets overall structure before attending to fine-grained elements. The final stage employs a multilayer perceptron (MLP) to map the integrated features into a concise quality score. Experimental results demonstrate that our method achieves improved consistency with human perceptual judgments compared to existing IQA models.

Figures

Figures reproduced from arXiv: 2504.17234 by the authors.

Figure 1
Figure 1. Qualitative comparison of different metrics against human preference on the BAPPS dataset. SPIPS consistently aligns [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The overall framework of our SPIPS model is structured into three modules: the traditional image quality assessment module (red), the low-level image perception feature assessment module (green), and the high-level image semantic feature assessment module (yellow). SPIPS takes two input images, Ieval (the image to be evaluated) and Iref (the reference or ground truth image). The traditional image quality assessment … view at source ↗
Figure 3
Figure 3. Qualitative comparison between the full SPIPS model and its ablated variant without the semantic module. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Qualitative comparison of the full SPIPS model and its ablated variant without traditional IQA metrics such as PSNR [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

54 extracted references · 49 canonical work pages

  1. [1]

    Simoncelli

    Ball ´e, Johannes, Valero Laparra, and Eero P. Simoncelli. ”End-to-end optimized image compression.” arXiv preprint arXiv:1611.01704 2016

    arXiv 2016

  2. [2]

    ”Variational image compression with a scale hyperprior.” arXiv preprint arXiv:1802.01436 2018

    Ball ´e, Johannes, David Minnen, Saurabh Singh, Sung Jin Hwang, and Nick Johnston. ”Variational image compression with a scale hyperprior.” arXiv preprint arXiv:1802.01436 2018

    arXiv 2018

  3. [3]

    Simoncelli, and Alan C

    Wang, Zhou, Eero P. Simoncelli, and Alan C. Bovik. ”Multiscale structural similarity for image quality assessment.” The Thrity-Seventh Asilomar Conference on Signals, Systems & Computers, 2003. V ol. 2. Ieee, 2003

    work page 2003

  4. [5]

    Efros, Eli Shechtman, and Oliver Wang

    Zhang, Richard, Phillip Isola, Alexei A. Efros, Eli Shechtman, and Oliver Wang. ”The unreasonable effectiveness of deep features as a perceptual metric.” In Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 586-595. 2018

    work page 2018

  5. [6]

    Kumar, Manoj, Neil Houlsby, Nal Kalchbrenner, and Ekin D. Cubuk. ”Do better ImageNet classifiers assess perceptual similarity better?.”, Transactions of Machine Le

    work page

  6. [7]

    Elements of information theory

    Cover, Thomas M. Elements of information theory. John Wiley & Sons, 1999

    work page 1999

  7. [8]

    ”What’s wrong with mean-squared error?.” In Digital images and human vision, pp

    Girod, Bernd. ”What’s wrong with mean-squared error?.” In Digital images and human vision, pp. 207-220. 1993

    work page 1993

  8. [10]

    Madhusudana, Pavan C., Neil Birkbeck, Yilin Wang, Balu Adsumilli, and Alan C. Bovik. ”Conviqt: Contrastive video quality estimator.” IEEE Transactions on Image Processing 32 (2023): 5138-5152

    work page 2023

Show all 54 references
  1. [11]

    ”Contrastive distortion-level learning-based no-reference image-quality assessment.” International Journal of Intelligent Systems 37, no

    Wei, Xuekai, Jing Li, Mingliang Zhou, and Xianmin Wang. ”Contrastive distortion-level learning-based no-reference image-quality assessment.” International Journal of Intelligent Systems 37, no. 11 (2022): 8730- 8746

    2022

  2. [12]

    ”Learned perceptual image enhancement.” In 2018 IEEE international conference on computational photography (ICCP), pp

    Talebi, Hossein, and Peyman Milanfar. ”Learned perceptual image enhancement.” In 2018 IEEE international conference on computational photography (ICCP), pp. 1-13. IEEE, 2018

    2018

  3. [14]

    ”Display device-adapted video quality-of-experience assessment.” In Human vision and electronic imaging XX, vol

    Rehman, Abdul, Kai Zeng, and Zhou Wang. ”Display device-adapted video quality-of-experience assessment.” In Human vision and electronic imaging XX, vol. 9394, pp. 27-37. SPIE, 2015

    2015

  4. [15]

    ”Gans trained by a two time-scale update rule converge to a local nash equilibrium.” Advances in neural information processing systems 30 2017

    Heusel, Martin, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. ”Gans trained by a two time-scale update rule converge to a local nash equilibrium.” Advances in neural information processing systems 30 2017

    2017

  5. [16]

    ”Improved techniques for training gans.” Advances in neural information processing systems 29 (2016)

    Salimans, Tim, Ian Goodfellow, Wojciech Zaremba, Vicki Cheung, Alec Radford, and Xi Chen. ”Improved techniques for training gans.” Advances in neural information processing systems 29 (2016)

    2016

  6. [17]

    Sutherland, Michael Arbel, and Arthur Gretton

    Bi ´nkowski, Mikołaj, Danica J. Sutherland, Michael Arbel, and Arthur Gretton. ”Demystifying mmd gans.” arXiv preprint arXiv:1801.01401 (2018)

    2018 arXiv

  7. [18]

    H. R. Sheikh, A. C. Bovik and G. de Veciana, ”An information fidelity criterion for image quality assessment using natural scene statistics,” in IEEE Transactions on Image Processing, vol. 14, no. 12, pp. 2117-2128, Dec. 2005

    2005

  8. [19]

    Simoncelli

    Ding, Keyan, Kede Ma, Shiqi Wang, and Eero P. Simoncelli. ”Image quality assessment: Unifying structure and texture similarity.” IEEE transactions on pattern analysis and machine intelligence 44, no. 5 (2020): 2567-2581

    2020

  9. [20]

    ”Visual quality assessment for super-resolved images: Database and method.” IEEE Transactions on Image Processing 28, no

    Zhou, Fei, Rongguo Yao, Bozhi Liu, and Guoping Qiu. ”Visual quality assessment for super-resolved images: Database and method.” IEEE Transactions on Image Processing 28, no. 7 (2019): 3528-3541

    2019

  10. [21]

    ”Perception-Driven Similarity-Clarity Tradeoff for Image Super-Resolution Quality Assessment.” IEEE Transactions on Circuits and Systems for Video Technology 34, no

    Zhang, Keke, Tiesong Zhao, Weiling Chen, Yuzhen Niu, Jinsong Hu, and Weisi Lin. ”Perception-Driven Similarity-Clarity Tradeoff for Image Super-Resolution Quality Assessment.” IEEE Transactions on Circuits and Systems for Video Technology 34, no. 7 (2023): 5897-5907

    2023

  11. [22]

    ”Convolutional neural networks for no-reference image quality assessment.” In Proceedings of the IEEE conference on computer vision and pattern recognition, pp

    Kang, Le, Peng Ye, Yi Li, and David Doermann. ”Convolutional neural networks for no-reference image quality assessment.” In Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 1733-1740. 2014

    2014

  12. [23]

    Madhusudana, Pavan C., Neil Birkbeck, Yilin Wang, Balu Adsumilli, and Alan C. Bovik. ”Image quality assessment using contrastive learn- ing.” IEEE Transactions on Image Processing 31 (2022): 4149-4161

    2022

  13. [24]

    ”DACNN: Blind image quality assessment via a distortion-aware convolutional neural network.” IEEE Transactions on Circuits and Systems for Video Technology 32, no

    Pan, Zhaoqing, Hao Zhang, Jianjun Lei, Yuming Fang, Xiao Shao, Nam Ling, and Sam Kwong. ”DACNN: Blind image quality assessment via a distortion-aware convolutional neural network.” IEEE Transactions on Circuits and Systems for Video Technology 32, no. 11 (2022): 7518- 7531

    2022

  14. [25]

    Zhou, Zehong, Fei Zhou, and Guoping Qiu. ”Blind image quality assessment based on separate representations and adaptive interaction of content and distortion.” IEEE Transactions on Circuits and Systems for Video Technology 34, no. 4 (2023): 2484-2497

    2023

  15. [26]

    Alireza, Saba Dadsetan, and Kris M

    Golestaneh, S. Alireza, Saba Dadsetan, and Kris M. Kitani. ”No- reference image quality assessment via transformers, relative ranking, and self-consistency.” In Proceedings of the IEEE/CVF winter confer- ence on applications of computer vision, pp. 1220-1230. 2022

    2022

  16. [27]

    ”Maniqa: Multi-dimension attention network for no-reference image quality assessment.” In Pro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp

    Yang, Sidi, Tianhe Wu, Shuwei Shi, Shanshan Lao, Yuan Gong, Ming- deng Cao, Jiahao Wang, and Yujiu Yang. ”Maniqa: Multi-dimension attention network for no-reference image quality assessment.” In Pro- ceedings of the IEEE/CVF conference on computer vision and pattern recognitio...

    2022

  17. [28]

    ”Data-efficient image quality assessment with attention- panel decoder.” In Proceedings of the AAAI Conference on Artificial Intelligence, vol

    Qin, Guanyi, Runze Hu, Yutao Liu, Xiawu Zheng, Haotian Liu, Xiu Li, and Yan Zhang. ”Data-efficient image quality assessment with attention- panel decoder.” In Proceedings of the AAAI Conference on Artificial Intelligence, vol. 37, no. 2, pp. 2091-2100. 2023

    2023

  18. [29]

    ”Blind quality assessment for in-the-wild images via hierarchical feature fusion and iterative mixed database training.” IEEE Journal of Selected Topics in Signal Processing 17, no

    Sun, Wei, Xiongkuo Min, Danyang Tu, Siwei Ma, and Guangtao Zhai. ”Blind quality assessment for in-the-wild images via hierarchical feature fusion and iterative mixed database training.” IEEE Journal of Selected Topics in Signal Processing 17, no. 6 (2023): 1178-1192

    2023

  19. [30]

    Digital image processing

    Gonzalez, Rafael C. Digital image processing. Pearson education india, 2009

    2009

  20. [31]

    ”Image quality metrics: PSNR vs

    Hore, Alain, and Djemel Ziou. ”Image quality metrics: PSNR vs. SSIM.” In 2010 20th international conference on pattern recognition, pp. 2366-

    2010

  21. [32]

    Bovik, Hamid R

    Wang, Zhou, Alan C. Bovik, Hamid R. Sheikh, and Eero P. Simoncelli. ”Image quality assessment: from error visibility to structural similarity.” IEEE transactions on image processing 13, no. 4 (2004): 600-612

    2004

  22. [34]

    ”Hadamard products and multivariate statistical analysis.” Linear algebra and its applications 6 (1973): 217-240

    Styan, George PH. ”Hadamard products and multivariate statistical analysis.” Linear algebra and its applications 6 (1973): 217-240

    1973

  23. [35]

    ”High-Fidelity 4x Neural Reconstruction of Real-time Path Traced Images.” In Proceedings of the Winter Conference on Applications of Computer Vision, pp

    Lao, Zhiqiang, Yu Guo, Xiyun Song, Yubin Zhou, Zongfang Lin, Heather Yu, and Liang Peng. ”High-Fidelity 4x Neural Reconstruction of Real-time Path Traced Images.” In Proceedings of the Winter Conference on Applications of Computer Vision, pp. 157-166. 2025

    2025

  24. [36]

    Lawrence Zitnick

    Lin, Tsung-Yi, Michael Maire, Serge Belongie, James Hays, Pietro Per- ona, Deva Ramanan, Piotr Doll ´ar, and C. Lawrence Zitnick. ”Microsoft coco: Common objects in context.” In Computer vision–ECCV 2014: 13th European conference, zurich, Switzerland, September 6-12, 2014, pro...

    2014

  25. [37]

    Wang, ShengYun Peng, Austin Wright, Kevin Li, Haekyu Park, Haoyang Yang, and Duen Horng Polo Chau

    Lee, Seongmin, Benjamin Hoover, Hendrik Strobelt, Zijie J. Wang, ShengYun Peng, Austin Wright, Kevin Li, Haekyu Park, Haoyang Yang, and Duen Horng Polo Chau. ”Diffusion explainer: Visual explanation for text-to-image stable diffusion.” In 2024 IEEE Visualization and Visual Ana...

    2024

  26. [38]

    ”A very preliminary analysis of DALL-E 2.” arXiv preprint arXiv:2204.13807 (2022)

    Marcus, Gary, Ernest Davis, and Scott Aaronson. ”A very preliminary analysis of DALL-E 2.” arXiv preprint arXiv:2204.13807 (2022)

    2022 arXiv

  27. [39]

    ”Glide: Towards photorealistic image generation and editing with text-guided diffusion models.” arXiv preprint arXiv:2112.10741 (2021)

    Nichol, Alex, Prafulla Dhariwal, Aditya Ramesh, Pranav Shyam, Pamela Mishkin, Bob McGrew, Ilya Sutskever, and Mark Chen. ”Glide: Towards photorealistic image generation and editing with text-guided diffusion models.” arXiv preprint arXiv:2112.10741 (2021)

    2021 arXiv

  28. [40]

    ”Improving image generation with better captions.” Computer Science

    Betker, James, Gabriel Goh, Li Jing, Tim Brooks, Jianfeng Wang, Linjie Li, Long Ouyang et al. ”Improving image generation with better captions.” Computer Science. https://cdn. openai. com/papers/dall-e-3. pdf 2, no. 3 (2023): 8

    2023

  29. [41]

    ”Pearson’s correlation coefficient.” Bmj 345 (2012)

    Sedgwick, Philip. ”Pearson’s correlation coefficient.” Bmj 345 (2012)

    2012

  30. [42]

    ”Spearman’s rank correlation coefficient.” Bmj 349 (2014)

    Sedgwick, Philip. ”Spearman’s rank correlation coefficient.” Bmj 349 (2014)

    2014

  31. [43]

    ”The Kendall rank correlation coefficient.” Encyclopedia of measurement and statistics 2 (2007): 508-510

    Abdi, Herv ´e. ”The Kendall rank correlation coefficient.” Encyclopedia of measurement and statistics 2 (2007): 508-510

    2007

  32. [44]

    Bovik, and Christophe Charrier

    Saad, Michele A., Alan C. Bovik, and Christophe Charrier. ”Blind image quality assessment: A natural scene statistics approach in the DCT domain.” IEEE transactions on Image Processing 21, no. 8 (2012): 3339- 3352

    2012

  33. [45]

    ”No- reference image quality assessment in the spatial domain.” IEEE Trans- actions on image processing 21, no

    Mittal, Anish, Anush Krishna Moorthy, and Alan Conrad Bovik. ”No- reference image quality assessment in the spatial domain.” IEEE Trans- actions on image processing 21, no. 12 (2012): 4695-4708

    2012

  34. [46]

    Ghadiyaram, Deepti, and Alan C. Bovik. ”Perceptual quality prediction on authentically distorted images using a bag of features approach.” Journal of vision 17, no. 1 (2017): 32-32

    2017

  35. [47]

    completely blind

    Mittal, Anish, Rajiv Soundararajan, and Alan C. Bovik. ”Making a “completely blind” image quality analyzer.” IEEE Signal processing letters 20, no. 3 (2012): 209-212

    2012

  36. [50]

    ”Topiq: A top-down approach from semantics to distortions for image quality assessment.” IEEE Transactions on Image Processing (2024)

    Chen, Chaofeng, 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 (2024)

    2024

  37. [51]

    Lao, Shanshan, Yuan Gong, Shuwei Shi, Sidi Yang, Tianhe Wu, Jiahao Wang, Weihao Xia, and Yujiu Yang. ”Attentions help cnns see better: Attention-based hybrid image quality assessment network.” In Proceed- ings of the IEEE/CVF conference on computer vision and pattern recogniti...

    2022

  38. [52]

    Sheikh, Hamid R., and Alan C. Bovik. ”Image information and visual quality.” IEEE Transactions on image processing 15, no. 2 (2006): 430- 444

    2006

  39. [53]

    ”A patch-structure representation method for quality assessment of contrast changed images.” IEEE Signal Processing Letters 22, no

    Wang, Shiqi, Kede Ma, Hojatollah Yeganeh, Zhou Wang, and Weisi Lin. ”A patch-structure representation method for quality assessment of contrast changed images.” IEEE Signal Processing Letters 22, no. 12 (2015): 2387-2390

    2015

  40. [54]

    Simoncelli, and Alan C

    Wang, Zhou, Eero P. Simoncelli, and Alan C. Bovik. ”Multiscale struc- tural similarity for image quality assessment.” In The Thrity-Seventh Asilomar Conference on Signals, Systems & Computers, 2003, vol. 2, pp. 1398-1402. Ieee, 2003

    2003

  41. [55]

    ”FSIM: A feature similarity index for image quality assessment.” IEEE transactions on Image Processing 20, no

    Zhang, Lin, Lei Zhang, Xuanqin Mou, and David Zhang. ”FSIM: A feature similarity index for image quality assessment.” IEEE transactions on Image Processing 20, no. 8 (2011): 2378-2386

    2011

  42. [56]

    ”VSI: A visual saliency-induced index for perceptual image quality assessment.” IEEE Transactions on Image processing 23, no

    Zhang, Lin, Ying Shen, and Hongyu Li. ”VSI: A visual saliency-induced index for perceptual image quality assessment.” IEEE Transactions on Image processing 23, no. 10 (2014): 4270-4281

    2014

  43. [57]

    ”Learning conditional knowledge distillation for degraded-reference im- age quality assessment.” In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp

    Zheng, Heliang, Huan Yang, Jianlong Fu, Zheng-Jun Zha, and Jiebo Luo. ”Learning conditional knowledge distillation for degraded-reference im- age quality assessment.” In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 10242-10251. 2021

    2021

  44. [58]

    ”Quality assessment of image super- resolution: Balancing deterministic and statistical fidelity.” In Proceed- ings of the 30th ACM international conference on multimedia, pp

    Zhou, Wei, and Zhou Wang. ”Quality assessment of image super- resolution: Balancing deterministic and statistical fidelity.” In Proceed- ings of the 30th ACM international conference on multimedia, pp. 934-

  45. [59]

    Zhang, Lin, Lei Zhang, and Alan C. Bovik. ”A feature-enriched com- pletely blind image quality evaluator.” IEEE Transactions on Image Processing 24, no. 8 (2015): 2579-2591

    2015

  46. [60]

    Sheikh, Hamid R., and Alan C. Bovik. ”A visual information fidelity approach to video quality assessment.” In The first international work- shop on video processing and quality metrics for consumer electronics, vol. 7, no. 2, pp. 2117-2128. sn, 2005

    2005

Pith tools

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