Pith. sign in

REVIEW 4 major objections 5 minor 26 references

Efficient Transformer for High Resolution Image Motion Deblurring

T0 review · 4 major / 5 minor · reviewed 2026-08-09 · deepseek-v4-flash

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

Pith's one-line read A 18.4% smaller Restormer transformer keeps deblurring quality while cutting training time by five hours.

desk verdict Honest reproduction study, but the efficiency claim is confounded by simultaneous changes to architecture, training schedule, augmentations, and loss. read the letter →

arxiv 2501.18403 v1 pith:B3EAOFUW submitted 2025-01-30 cs.CV cs.AI

Amanturdieva Akmaral , Muhammad Hamza Zafar This is my paper · ORCID
classification cs.CVcs.AI MSC 68T0768U10
keywords imagedeblurringRestormertransformermotionblurefficientattentionfrequencylossdataaugmentationhigh-resolutionimages
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

This paper argues that the Restormer transformer, a general-purpose image restoration network, is over-parameterized for motion deblurring. By pruning roughly 30% of its layers and transformer blocks while doubling the number of attention heads per stage, the authors obtain a model with 18.4% fewer parameters that trains about five hours faster and matches or slightly exceeds the original's PSNR and SSIM on real-world blur benchmarks. They attribute part of the gain to an enriched training pipeline—color jitter, Gaussian blur, perspective transforms, and a Fourier magnitude loss—that smooths convergence. The result matters because it suggests that careful architectural simplification, not just bigger models, can make transformer-based deblurring practical on limited hardware.

What carries the argument

The core mechanism is the interplay between two Restormer components—Multi-Dconv Head Transposed Attention (MDTA), which computes channel-wise attention with linear complexity, and the Gated-Dconv Feed-Forward Network (GDFN). The modification reduces the number of these blocks across the network but doubles the attention heads per MDTA stage, so the model attends to more subspaces in fewer layers. A Fourier magnitude loss, $L_{freq} = \frac{1}{N}\sum_i\||F(\hat{I}_i)| - |F(I_i)|\|_1$ with weighting $\lambda=0.1$, is added to the pixel L1 loss to preserve high-frequency detail, and the augmented training pipeline (color jitter, Gaussian blur, brightness/contrast, perspective transforms) is intended to bridge the color and geometric variability of real blurred images.

What would settle it

Run a controlled comparison in which the original Restormer and the simplified model are trained from scratch with identical data splits, augmentations, loss, and the same progressive patch/batch schedule; if the simplified model's PSNR and SSIM then fall below the original's on RealBlur-R or UHDM, the claimed 'maintaining or improving performance' is an artifact of the training recipe rather than the architectural change.

Watch

Extended reading notes

Core claim

The central claim is that a deliberately simplified Restormer variant can maintain or improve motion-deblurring quality while being substantially cheaper. The authors reduce the number of transformer blocks and layers in the encoder, latent, decoder, and refinement stages, doubling the attention heads per MDTA stage to preserve representational capacity, and combine this with extra data augmentations and a frequency-domain loss. On RealBlur-R, the improved model reaches 33.997 dB PSNR versus 33.685 dB for their reproduced baseline; on UHDM it reaches 21.359 dB versus 21.260 dB; and it cuts both training time (23 versus 28 hours) and inference time. The improved model also lowers the count of hard negative examples on UHDM from 604 to 598. The authors present this as evidence that many existing architectures may be over-parameterized for their target tasks.

Load-bearing premise

The evaluation assumes that the reproduced baseline and the improved model differ only by the intended modifications, but the baseline was trained with a smaller progressive batch schedule and the RealBlur train/validation split is not known, so the reported performance differences could stem from these uncontrolled factors rather than the architecture.

Editorial extensions

If this is right

  • The 18.4% parameter and 30% layer reductions shrink the model to 81.5 MB versus 99.9 MB, making transformer deblurring more feasible on memory-limited devices.
  • Faster convergence and smoother loss curves suggest the simplified architecture is easier to train, lowering the compute budget needed to reach a given quality.
  • The improved model's higher PSNR and fewer hard negatives on UHDM indicate the simplification generalizes to very high-resolution blur rather than only the training distribution.
  • Fine-tuned models on RealBlur degrade sharply on UHDM, revealing a specialization–overfitting tradeoff that the improved model avoids.
  • The combined architectural and training changes yield a net win even though color augmentation alone did not improve color fidelity on UHDM, pointing to the architecture as the main efficiency driver.

Reading between the lines

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

  • The uncontrolled differences in progressive-training schedules between the reproduced baseline and the improved model mean the claimed architectural benefit is not cleanly isolated; a matched-schedule comparison could still confirm it, but the current evidence conflates architecture with training recipe.
  • If the simplification principle transfers to other restoration tasks such as denoising and super-resolution, it would imply that Restormer's original block counts are not a universal optimum and task-specific pruning could be a general recipe.
  • The Fourier magnitude loss is a lightweight addition that could be ported to other restoration architectures, but its contribution is not separated from the augmentations in the ablation, so a factorial study would reveal which component drives the smoother convergence.
  • The UHDM results, where all models score below 21.4 dB PSNR, suggest that high-resolution synthetic blur remains a hard target; a worthwhile extension would test the simplified model on real 4K–6K motion blur rather than only kernel-convolved sharp images.
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 / 5 minor

Summary. This paper presents an empirical study of Restormer for high-resolution image motion deblurring. The authors reproduce a baseline Restormer, fine-tune the authors' checkpoint on RealBlur-R/J, and propose an 'improved' variant that reduces the number of transformer blocks/layers and total parameters, doubles the attention heads per stage, adds color jitter/Gaussian blur/brightness-contrast/perspective augmentations, and trains with a composite loss consisting of pixel L1 plus a Fourier-magnitude loss. The paper reports PSNR, SSIM, MAE, LPIPS, and DeltaE on RealBlur-R, RealBlur-J, and UHDM, plus training and inference time comparisons. The main claim is that the improved model maintains or improves deblurring performance while reducing model complexity by 18.4% and total layers by 30%.

Significance. If the central claim were established, the paper would provide a practically useful data point: a smaller Restormer variant with comparable high-resolution deblurring quality and lower computational cost. The authors are transparent about their training schedules and about the difficulties of reproducing the original Restormer results, which is a strength. However, the evidence does not currently isolate the architectural modification as the cause of the observed performance; the improved model differs from the baseline simultaneously in architecture, training schedule, augmentations, and loss function, and the reported performance differences are small relative to plausible training variability. The paper also provides no code or trained model in the submission beyond an unverified repository URL, so the reproducibility of the claimed 18.4% parameter reduction depends entirely on a precise architectural description that is not fully given in the text.

major comments (4)
  1. [Section VII; Section VI.B] The central efficiency claim is not supported because the comparison is confounded. The reproduced baseline (Section VI.B) is trained with the progressive schedule {(128,8),(160,4),(192,4),(256,2),(320,1),(320,1)}, L1 loss only, and no color or geometric augmentations beyond flips, whereas the improved model (Section VII) is trained with {(128,8),(160,6),(192,4),(256,2),(320,2),(384,1)}, the composite loss L_total = L_pixel + 0.1 L_freq, and additional color jitter, Gaussian blur, brightness/contrast, and perspective transforms. The 'ablation studies' in Section VII compare the full changed system against the reproduced baseline; no experiment varies the architecture alone. Therefore the maintenance or improvement in Tables III, V, and VI cannot be attributed to the 18.4% parameter reduction and 30% layer reduction claimed in the abstract.
  2. [Section VI.D; Tables III, V, VI] The quantitative evidence for improvement is statistically weak. All reported values come from a single training run, with no error bars, multiple seeds, or significance tests. On UHDM the improved model's PSNR is 21.359 dB versus 21.323 dB for the provided checkpoint and 21.260 dB for the reproduced model; on RealBlur-R the improved model (33.997 dB) is essentially identical to the checkpoint (33.998 dB). Differences of roughly 0.1 dB on a single seed are within the range of random initialization and training variability, especially given the paper's own acknowledgment that data-split details are unknown. The authors should report means and standard deviations over multiple seeds, or a paired significance test, before claiming that the improved model maintains or improves performance.
  3. [Section VI.B; Tables II and IV] The comparison against the published Restormer results is not valid as presented. The rows labeled 'Results in the paper' were obtained with 8 GPUs, batch sizes up to 64, and the schedule {(160,40),(192,32),(256,16),(320,8),(384,8)}, while the reproduced model is trained on 1 GPU with batch sizes up to 8 and a different schedule. Section VI.B itself states that lack of information about the train/validation split could explain the deviations. Consequently, neither the reproduced nor the improved model can be said to 'maintain' the original Restormer's performance; the comparison conflates architecture with training protocol and data handling.
  4. [Section IV; Figures 2 and 3] The architectural modification is not specified with enough precision to verify the claimed parameter reduction. The text states that the number of transformer blocks and layers was decreased and that attention heads per stage were doubled, but it does not give the exact per-level block counts, head counts, channel counts, or the refinement-stage configuration of the improved model. Without these numeric details, the 18.4% parameter reduction and 30% layer reduction cannot be independently reproduced from the manuscript, and the efficiency claim is not checkable.
minor comments (5)
  1. [Table I] The text refers to a hard-example threshold of 'PSNR between 20dB and 30dB', but Table I says 'PSNR between 20dB and 3dB'; the definition of Hard Positives/Hard Negatives is also not formalized, and the counts are not obviously comparable across datasets of different sizes.
  2. [Section VI.A] The enumeration of the four experimental configurations is inconsistent: the word 'Second' is used twice and the third/fourth configurations are not clearly introduced, making the experimental setup harder to follow.
  3. [Section VII] The phrase 'detailed ablation studies' overstates the content: the section compares the full improved system with the reproduced baseline and shows training curves, but it does not ablate the frequency-loss weight, the augmentation components, or the architectural changes individually.
  4. [Table VI; Discussion] On UHDM the improved model has a worse DeltaE (4.023) than the reproduced baseline (3.867), which the paper acknowledges in passing. The abstract's blanket statement that the model 'maintains or improves performance' should be qualified to reflect that color fidelity is not improved on this benchmark.
  5. [General] The GitHub link is mentioned but the manuscript does not state whether code, model weights, or evaluation scripts are included, nor does it provide a versioned release or license, which weakens the reproducibility statement.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: results come from supervised training against external benchmarks, and no fitted quantity is reused as evidence or prediction.

full rationale

The paper's central claim is that an architectural modification of Restormer reduces parameters by 18.4% and layers by 30% while maintaining deblurring performance, supported by PSNR, SSIM, MAE, LPIPS, and DeltaE measurements on RealBlur-R, RealBlur-J, and UHDM. These are external, supervised benchmark evaluations; the training loss (L1 plus a Fourier magnitude term with weight 0.1) is not simultaneously used to define any evaluation metric, so no fitted hyperparameter is renamed as a prediction. The comparison is confounded because the reproduced baseline and the improved model differ in architecture, progressive training schedule, augmentations, and loss function simultaneously, but confounding is an experimental-validity problem, not circularity. The paper does not invoke a uniqueness theorem, does not rely on a load-bearing self-citation chain, and does not define any claimed result in terms of itself. The only citations to prior work are standard external sources such as Restormer, RealBlur, and UHDM, and no central premise is justified solely by the present authors' own prior publications. Therefore no circular step is identified, and the analysis should focus on confounds and reproducibility rather than circular reasoning.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim depends on supervised deep learning with external datasets. The listed free parameters are hand-chosen training and analysis choices; no invented entities are introduced. The axioms are standard domain assumptions about metrics and transfer, but they are not verified by the paper.

free parameters (5)
  • Frequency loss weight lambda = 0.1
    Set by hand in Section IV; balances L_pixel and L_freq; not optimized, but affects the training objective.
  • Attention head multiplier for improved model = 2x heads per stage
    Chosen empirically to compensate for fewer blocks; exact number of heads after the change is not stated.
  • Layer/block reduction factor = 30% fewer layers and 18.4% fewer parameters
    Selected by the authors; exact new block counts per stage are not given, so the design choice is not reproducible from the text.
  • Augmentation policy = color jitter, Gaussian blur, brightness/contrast, perspective transforms
    Hand-selected to improve robustness; no hyperparameter ranges or strengths are given.
  • Hard example PSNR threshold = 20-30 dB (text has typo '3dB')
    Used in Table I to define hard positives and negatives; threshold chosen post hoc and not justified.
assumptions (4)
  • domain assumption PSNR, SSIM, LPIPS, and deltaE2000 are adequate proxies for deblurring quality.
    Used throughout the evaluation without justification that these metrics capture the claimed perceptual improvements.
  • domain assumption Models trained on GoPro or RealBlur transfer meaningfully to UHDM despite large resolution and blur kernel differences.
    The paper evaluates on UHDM without fine-tuning and ranks variants by those cross-domain numbers.
  • domain assumption The RealBlur benchmark split used for fine-tuning and evaluation is representative.
    Section VI.B admits the exact train/validation split used by the original authors is unknown, so the reproduced baseline may be evaluated on a different split.
  • domain assumption A Fourier magnitude difference is a valid frequency-domain loss for deblurring.
    The frequency loss in Section IV is adopted from prior work [1] without a derivation or analysis of its failure modes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Efficient Transformer for High Resolution Image Motion Deblurring." pith.science (2026). https://pith.science/paper/B3EAOFUW

@misc{pith2026250118403,
  author       = {Pith},
  title        = {Pith review of: Efficient Transformer for High Resolution Image Motion Deblurring},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/B3EAOFUW}},
  note         = {Machine review of arXiv:2501.18403}
}
read the original abstract

This paper presents a comprehensive study and improvement of the Restormer architecture for high-resolution image motion deblurring. We introduce architectural modifications that reduce model complexity by 18.4% while maintaining or improving performance through optimized attention mechanisms. Our enhanced training pipeline incorporates additional transformations including color jitter, Gaussian blur, and perspective transforms to improve model robustness as well as a new frequency loss term. Extensive experiments on the RealBlur-R, RealBlur-J, and Ultra-High-Definition Motion blurred (UHDM) datasets demonstrate the effectiveness of our approach. The improved architecture shows better convergence behavior and reduced training time while maintaining competitive performance across challenging scenarios. We also provide detailed ablation studies analyzing the impact of our modifications on model behavior and performance. Our results suggest that thoughtful architectural simplification combined with enhanced training strategies can yield more efficient yet equally capable models for motion deblurring tasks. Code and Data Available at: https://github.com/hamzafer/image-deblurring

Figures

Figures reproduced from arXiv: 2501.18403 by the authors.

Figure 1
Figure 1. Architecture of Restormer for high-resolution image restoration from [20] (a) Multi-Dconv Head Transposed Attention [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Detailed architecture stage comparison across encoder, [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Model architecture changes showing reduction in [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (14 more)
Figure 6
Figure 6. Figure 6: Tensorboard training progression of PSNR over 300K [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Tensorboard training progression of PSNR over 300K [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 5
Figure 5. Figure 5: Tensorboard training progression of L1 loss over 300K [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 8
Figure 8. Figure 8: Tensorboard training progression of SSIM over 300K [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]
Figure 9
Figure 9. Figure 9: Tensorboard training progression of SSIM over 300K [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]
Figure 10
Figure 10. Figure 10: GoPro [9] B. ReaBlur Dataset Examples in [PITH_FULL_IMAGE:figures/full_fig_p011_10.png]
Figure 11
Figure 11. Figure 11: RealBlur [10] [PITH_FULL_IMAGE:figures/full_fig_p011_11.png]
Figure 12
Figure 12. Figure 12: UHDM [23] D. RealBlur-R Dataset Examples Fig.13 shows a hard negative example from the RealBlur-R dataset, while Fig.14 presents a hard positive example from the same dataset. These examples demonstrate the subtle differences that make classification challenging in re…
Figure 13
Figure 13. Figure 13: Hard Negative example from RealBlur-J dataset (zoom in for better visibility) [PITH_FULL_IMAGE:figures/full_fig_p013_13.png]
Figure 14
Figure 14. Figure 14: Hard Positive example from RealBlur-J dataset (zoom in for better visibility) [PITH_FULL_IMAGE:figures/full_fig_p013_14.png]
Figure 15
Figure 15. Figure 15: Hard Negative example from RealBlur-J [10] dataset (zoom in for better visibility) [PITH_FULL_IMAGE:figures/full_fig_p013_15.png]
Figure 16
Figure 16. Figure 16: Hard Positive example from RealBlur-J [10] dataset (zoom in for better visibility) [PITH_FULL_IMAGE:figures/full_fig_p014_16.png]
Figure 17
Figure 17. Figure 17: Hard Negative example from UHDM [23] dataset (zoom in for better visibility) [PITH_FULL_IMAGE:figures/full_fig_p014_17.png]
Figure 18
Figure 18. Figure 18: Hard Positive example from UHDM [23] dataset (zoom in for better visibility) [PITH_FULL_IMAGE:figures/full_fig_p014_18.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 17 canonical work pages

  1. [1]

    Guided Frequency Loss for Image Restoration

    Bilel Benjdira, Anas M. Ali, and Anis Koubaa. Guided fre- quency loss for image restoration. (arXiv:2309.15563), October 2023. arXiv:2309.15563 [cs, eess]

    work page Pith review arXiv 2023

  2. [2]

    A Comprehensive Survey on Deep Neural Image Deblurring

    Sajjad Amrollahi Biyouki and Hoon Hwangbo. A Comprehen- sive Survey on Deep Neural Image Deblurring, October 2023. arXiv:2310.04719

    work page Pith review arXiv 2023

  3. [3]

    Improving Image Restoration by Revisiting Global Information Aggregation

    Xiaojie Chu, Liangyu Chen, Chengpeng Chen, and Xin Lu. Improv- ing Image Restoration by Revisiting Global Information Aggregation, August 2022. arXiv:2112.04491 [cs]

    work page Pith review arXiv 2022

  4. [4]

    An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale, June 2021

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weis- senborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, and Neil Houlsby. An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale, June 2021. arXiv:2010.11929

    arXiv 2021

  5. [5]

    Gaussian Error Linear Units (GELUs), June 2016

    Dan Hendrycks and Kevin Gimpel. Gaussian Error Linear Units (GELUs), June 2016. arXiv:1606.08415

    arXiv 2016

  6. [6]

    Real-Time Motion Blur Using Multi-Layer Motion Vectors

    Donghyun Lee, Hyeoksu Kwon, and Kyoungsu Oh. Real-Time Motion Blur Using Multi-Layer Motion Vectors. Applied Sciences, 14(11):4626, January 2024. Number: 11 Publisher: Multidisciplinary Digital Publish- ing Institute

    work page 2024

  7. [7]

    Decoupled Weight Decay Regular- ization, November 2017

    Ilya Loshchilov and Frank Hutter. Decoupled Weight Decay Regular- ization, November 2017. arXiv:1711.05101

    arXiv 2017

  8. [8]

    SGDR: Stochastic Gradient Descent with Warm Restarts, May 2017

    Ilya Loshchilov and Frank Hutter. SGDR: Stochastic Gradient Descent with Warm Restarts, May 2017. arXiv:1608.03983 [cs]

    arXiv 2017

Show all 26 references
  1. [9]

    Deep multi-scale convolutional neural network for dynamic scene deblurring

    Seungjun Nah, Tae Hyun Kim, and Kyoung Mu Lee. Deep multi-scale convolutional neural network for dynamic scene deblurring. In 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , page 257–265, Honolulu, HI, July 2017. IEEE

    2017

  2. [10]

    Real- World Blur Dataset for Learning and Benchmarking Deblurring Al- gorithms

    Jaesung Rim, Haeyun Lee, Jucheol Won, and Sunghyun Cho. Real- World Blur Dataset for Learning and Benchmarking Deblurring Al- gorithms. In Andrea Vedaldi, Horst Bischof, Thomas Brox, and Jan- Michael Frahm, editors, Computer Vision – ECCV 2020 , volume 12370, pages 184–201. Sp...

    2020

  3. [11]

    U-Net: Con- volutional Networks for Biomedical Image Segmentation, May 2015

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-Net: Con- volutional Networks for Biomedical Image Segmentation, May 2015. arXiv:1505.04597

    2015 arXiv

  4. [12]

    Color-Sensitivity-Based Combined PSNR for Objective Video Quality Assessment

    Xiwu Shang, Jie Liang, Guozhong Wang, Haiwu Zhao, Chengjia Wu, and Chang Lin. Color-Sensitivity-Based Combined PSNR for Objective Video Quality Assessment. IEEE Transactions on Circuits and Systems for Video Technology, 29(5):1239–1250, May 2019. Conference Name: IEEE Transact...

    2019

  5. [13]

    Gaurav Sharma, Wencheng Wu, and Edul N. Dalal. The ciede2000 color- difference formula: Implementation notes, supplementary test data, and mathematical observations. Color Research & Application , 30(1):21–30, February 2005

    2005

  6. [14]

    Aitken, Rob Bishop, Daniel Rueckert, and Zehan Wang

    Wenzhe Shi, Jose Caballero, Ferenc Husz ´ar, Johannes Totz, Andrew P. Aitken, Rob Bishop, Daniel Rueckert, and Zehan Wang. Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network. In 2016 IEEE Conference on Computer Vision an...

    2016

  7. [15]

    Stripformer: Strip transformer for fast image deblurring

    Fu-Jen Tsai, Yan-Tsung Peng, Yen-Yu Lin, Chung-Chi Tsai, and Chia- Wen Lin. Stripformer: Strip transformer for fast image deblurring. In Shai Avidan, Gabriel Brostow, Moustapha Ciss ´e, Giovanni Maria Farinella, and Tal Hassner, editors, Computer Vision – ECCV 2022 , page 146–...

    2022

  8. [16]

    Gomez, Lukasz Kaiser, and Illia Polosukhin

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. Attention Is All You Need, June 2017. arXiv:1706.03762

    2017 arXiv

  9. [17]

    Uformer: A General U-Shaped Transformer for Image Restoration

    Zhendong Wang, Xiaodong Cun, Jianmin Bao, Wengang Zhou, Jianzhuang Liu, and Houqiang Li. Uformer: A General U-Shaped Transformer for Image Restoration. In 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages 17662–17672, New Orleans, LA, USA, June...

    2022

  10. [18]

    Bovik, H.R

    Zhou Wang, A.C. Bovik, H.R. Sheikh, and E.P. Simoncelli. Image quality assessment: from error visibility to structural similarity. IEEE Transactions on Image Processing , 13(4):600–612, April 2004. Confer- ence Name: IEEE Transactions on Image Processing

    2004

  11. [19]

    Sou Yoshihara, Taiki Fukiage, and Shin’ya Nishida. Does training with blurred images bring convolutional neural networks closer to humans with respect to robust object recognition and internal representations? Frontiers in Psychology , 14:1047694, February 2023

    2023

  12. [20]

    Restormer: Efficient Transformer for High-Resolution Image Restoration, March 2022

    Syed Waqas Zamir, Aditya Arora, Salman Khan, Munawar Hayat, Fahad Shahbaz Khan, and Ming-Hsuan Yang. Restormer: Efficient Transformer for High-Resolution Image Restoration, March 2022. arXiv:2111.09881

    2022 arXiv

  13. [21]

    Multi-Stage Progressive Image Restoration, March 2021

    Syed Waqas Zamir, Aditya Arora, Salman Khan, Munawar Hayat, Fahad Shahbaz Khan, Ming-Hsuan Yang, and Ling Shao. Multi-Stage Progressive Image Restoration, March 2021. arXiv:2102.02808

    2021 arXiv

  14. [22]

    Deep Image Deblurring: A Survey, May 2022

    Kaihao Zhang, Wenqi Ren, Wenhan Luo, Wei-Sheng Lai, Bjorn Stenger, Ming-Hsuan Yang, and Hongdong Li. Deep Image Deblurring: A Survey, May 2022. arXiv:2201.10700

    2022 arXiv

  15. [23]

    MC- Blur: A Comprehensive Benchmark for Image Deblurring, September

    Kaihao Zhang, Tao Wang, Wenhan Luo, Boheng Chen, Wenqi Ren, Bjorn Stenger, Wei Liu, Hongdong Li, and Ming-Hsuan Yang. MC- Blur: A Comprehensive Benchmark for Image Deblurring, September

  16. [24]

    Efros, Eli Shechtman, and Oliver Wang

    Richard Zhang, Phillip Isola, Alexei A. Efros, Eli Shechtman, and Oliver Wang. The Unreasonable Effectiveness of Deep Features as a Perceptual Metric, April 2018. arXiv:1801.03924

    2018 arXiv

  17. [25]

    United defocus blur detection and deblurring via adversarial promoting learning

    Wenda Zhao, Fei Wei, You He, and Huchuan Lu. United defocus blur detection and deblurring via adversarial promoting learning. In Shai Avidan, Gabriel Brostow, Moustapha Ciss ´e, Giovanni Maria Farinella, and Tal Hassner, editors, Computer Vision – ECCV 2022 , page 569–586, Cha...

    2022

  18. [26]

    UHD Image Deblurring via Multi-scale Cubic-Mixer, June 2022

    Zhuoran Zheng and Xiuyi Jia. UHD Image Deblurring via Multi-scale Cubic-Mixer, June 2022. arXiv:2206.03678 [cs]. APPENDIX This appendix presents remaining results and examples from three different datasets: RealBlur-R, RealBlur-J, and UHDM are provided. For each dataset, we al...

    2022 arXiv

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.