Pith. sign in

REVIEW 4 major objections 6 minor 112 references

Global Modeling Matters: A Fast, Lightweight and Effective Baseline for Efficient Image Restoration

T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A pyramid wavelet-Fourier network replaces self-attention with frequency-domain token mixing and reports state-of-the-art quality on seven restoration tasks at a fraction of the parameter and latency costs of current transformer and Mamba…

desk verdict A broad, well-executed empirical study of a lightweight wavelet-FFT restoration net whose quality claims look credible but whose efficiency claims need re-measurement before I'd trust them. read the letter →

arxiv 2507.13663 v1 pith:R3ESOQSE submitted 2025-07-18 cs.CV

Xingyu Jiang , Ning Gao , Hongkun Dou , Xiuhui Zhang , Xiaoqing Zhong , Yue Deng , Hongjue Li This is my paper
classification cs.CV
keywords imagerestorationwavelettransformFourierglobalmodelingefficientbaselinederainingsuper-resolutionmotiondeblurring
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 tries to establish that the global modeling power credited to transformers in image restoration does not require self-attention at all: a carefully placed combination of wavelet decomposition and Fourier transforms can deliver the same or better restoration quality with far less complexity. It proposes PW-FNet, a network whose inter-block design is a pyramid wavelet multi-input multi-output structure and whose intra-block token mixer is a global Fourier transform, and reports top PSNR/SSIM on deraining, raindrop removal, super-resolution, deblurring, dehazing, desnowing, underwater and low-light tasks. The practical significance is that current state-of-the-art restoration models are too large and slow for real-time deployment, and the paper argues that a lightweight baseline with 0.7 to 1.4 million parameters can serve the same role. If the claim holds, the field gets a cheap default architecture and a concrete argument that research effort should go into what global modeling is for, not into the attention mechanism itself.

What carries the argument

The load-bearing machinery is a two-level wavelet-Fourier design. Inter-block, pyramid wavelet multi-input multi-output (PW-MIMO) applies wavelet transforms to the input image to produce multi-scale lossless pyramid inputs $\{I^1, I^2, I^4\}$, passes them through a three-level encoder-decoder with trainable wavelet downsampling and upsampling, and emits three restored outputs so that small, medium, and large inference variants come from one set of weights. Intra-block, the token mixer of each PW-FNet block is a global 2D Fourier transform, pointwise convolutions, GELU, and an inverse Fourier transform, formalized as $f_5 = \mathcal{F}^{-1}(\mathrm{GELU}(\mathcal{F}(f_1 * W_{1\times1}) * W_{1\times1}) * W_{1\times1})$, which replaces self-attention; the feed-forward part retains a depthwise convolution for local refinement. The loss is taken in the Fourier domain, $L = \sum_i \lVert \mathcal{F}(o_i) - \mathcal{F}(g_i)\rVert_1$, which the ablations show works better than spatial, wavelet, or combined losses. The argument is carried by the paper's earlier observation, made without any learned network, that rain and similar degradations populate the high-frequency regions of the LL and HL sub-bands, so the architecture should narrow its feature-processing space there.

What would settle it

Run every method compared in Tables I, III, IV, V, VII and XIV from released checkpoints on one machine under one protocol (same GPU, batch size, input resolution, precision and inference framework), and record PSNR/SSIM, memory, FLOPs and wall-clock latency. If PW-FNet no longer leads on both quality and efficiency, or if its latency margin over Restormer, MPRNet and MaIR shrinks to a small fraction of the reported 27.7ms vs 85.7ms, 28.6ms vs 1143.2ms and 28.6ms vs 996.8ms, the paper's central claim is not supported.

Watch

Extended reading notes

Core claim

The paper's central claim is that degradation in natural images concentrates in compact high-frequency regions of a few wavelet sub-bands, and that a network built to exploit this can replace self-attention with global Fourier transforms without losing modeling power. Concretely, the paper claims PW-FNet-L reaches 42.23 dB PSNR on Rain200L with 1.44 million parameters and 33.56G FLOPs, where the previous state-of-the-art NeRD-Rain reaches 41.71 dB with 22.89 million parameters and 156.3G FLOPs; similar margins are reported for high-resolution deraining (35.93 dB vs 34.30 dB for UDR-Mixer on 4K-Rain13K), motion deblurring (34.03 dB on GoPro), efficient super-resolution, dehazing, desnowing, underwater and low-light enhancement. It also claims that one trained network can be executed as small, medium, or large variants by reading outputs at different pyramid stages, and that training with a Fourier-domain L1 loss alone outperforms spatial or wavelet losses in its ablations.

Load-bearing premise

The central claim depends on the cross-paper comparisons being fair: baseline quality scores are taken from their original papers, while memory, FLOPs, and latency are measured by the authors at different resolutions and on their own hardware, so a unified re-measurement could shrink the reported efficiency margins.

Editorial extensions

If this is right

  • Efficient image restoration would no longer need attention or state-space tokens: a frequency-domain token mixer plus wavelet decomposition could supply the global context transformers are prized for.
  • One trained PW-FNet would yield small, medium, and large models from the same weights, letting deployments trade accuracy for speed without retraining.
  • The reported latencies imply real-time high-resolution denoising and deraining: PW-FNet-S runs at 39ms per 1024x1024 frame on 4K-Rain13K, where transformer baselines take 547ms to 2682ms.
  • The ablation shows global FFT outperforms windowed Fourier kernels of sizes 8x8 to 64x64, suggesting the global receptive field itself, rather than locality, carries much of the restoration benefit.
  • A Fourier-domain loss alone can outperform spatial, wavelet, and combined losses, pointing to frequency supervision as a natural match for frequency-based token mixing.

Reading between the lines

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

  • The paper does not ablate the wavelet pyramid away while keeping the Fourier mixer, so a natural follow-up test is whether a plain single-scale network with global FFT token mixing retains most of the gain; that experiment would isolate whether the multi-frequency decomposition or the global Fourier operator is the active ingredient.
  • Because the degradation-localization story is demonstrated for rain with a hand-designed swapping pipeline, the same high-frequency-localization logic could be checked quantitatively for blur, haze, snow and low light by measuring sub-band energy differences between degraded and clean image pairs; the paper reports the rain case but not the others.
  • The reported latency advantages are measured under the authors' own protocol, so a deployment-oriented benchmark that sweeps batch size, precision, and hardware would be a direct test of whether the efficiency claim survives outside the paper's setup.
  • If the frequency-domain token mixer generalizes as claimed, one shared pretrained PW-FNet backbone could plausibly be fine-tuned across degradation types or serve as a common initialization for task-specific heads, a direction the paper does not explore.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes PW-FNet, a lightweight image-restoration network whose inter-block design is a pyramid wavelet multi-input multi-output (PW-MIMO) structure and whose intra-block design replaces self-attention with global Fourier transforms. The authors first present a wavelet-Fourier iterative pipeline as motivation, then evaluate PW-FNet on deraining, raindrop removal, super-resolution, motion deblurring, dehazing, desnowing, and underwater/low-light enhancement. They report state-of-the-art or competitive PSNR/SSIM with substantially fewer parameters, FLOPs, and lower latency than many transformer- and Mamba-based baselines, and they include ablations on the MIMO structure, wavelet family, Fourier kernel size, and loss function.

Significance. If the reported quality and efficiency numbers hold, PW-FNet would be a genuinely useful baseline: the Rain200L result of 42.23 dB with 1.442M parameters and 33.56G FLOPs compares favorably with much larger models, and the breadth of tasks tested is a strength. The paper also ships a public code link and includes informative ablations (Tables X-XIV) that isolate the contributions of the wavelet pyramid, global FFT, and Fourier-domain loss. The quality comparisons are not circular because they are made against independent external methods on standard benchmarks. However, the paper's headline efficiency claim rests on FLOPs and latency measurements whose counting method and evaluation protocol are not fully specified, and one loss term is undefined; these issues need to be resolved before the central claim can be taken at face value.

major comments (4)
  1. [Section IV.B, Table I vs. Section IV.H, Table IX; Fig.1 caption] The FLOPs accounting for PW-FNet is not stated, and the reported numbers contain an unexplained scaling inconsistency. The Fig.1 caption specifies 256x256 for the figure, and Table I lists PW-FNet-S at 16.64G FLOPs; Table IX lists the same 0.72M-parameter model at 55.6G FLOPs at 512x512. Doubling spatial resolution should roughly quadruple FLOPs, giving about 66.6G, not 55.6G. This ~16% shortfall suggests different counting conditions, resolutions, or model configurations across tables. More fundamentally, the paper never describes how the global 2D FFT and wavelet transforms in the PW-FNet block are counted; if these are non-convolutional operations omitted by the FLOPs tool, the computational cost of the architecture is understated. Please state the counting tool and method, and re-verify all FLOPs tables under a single, consistently documented protocol.
  2. [Section III.D, Eq. (3)] The Fourier-domain loss uses ground-truth quantities {g^i, i=1,2,4} that are never defined. The outputs {o^i} are defined as {I^i + r^i} with wavelet-transformed inputs {I^i}, but no similar definition is given for the corresponding {g^i}; it is unclear whether they are wavelet-transformed ground-truth images, Fourier spectra, or something else. Without this definition, Eq. (3) is not reproducible, and the ablation in Table XIII comparing spatial, wavelet, and Fourier losses cannot be fully interpreted.
  3. [Section I and Fig.2] The motivating claim that the pyramid wavelet-Fourier pipeline 'quantitatively' demonstrates where degradation concentrates is not supported by any numeric evidence in the text. The caption mentions quantitative results of swapping different sub-bands in the bottom-right corner of Fig.2, but the paper does not report those values or explain the swap protocol. Since this pipeline is offered as the design rationale for PW-MIMO and the Fourier block, the reader cannot verify the claimed observations; please add the numerical table or a clear reference to an appendix.
  4. [Section IV.F, Tables VII and IX; Section IV.B, Table III] The latency and memory comparisons that support the abstract's 'significantly reduced inference time' claim are reported without a common evaluation protocol. Tables VII and IX give memory and latency for PW-FNet and baselines but do not specify the GPU model, software versions, precision (FP32 vs. FP16), batch size, warmup iterations, or whether all methods are measured under identical conditions. Table III likewise reports latency for high-resolution deraining without such details. Cross-paper efficiency comparisons can shrink or reverse when re-measured under a single protocol, so please either provide the complete measurement setup for all efficiency tables or explicitly qualify the efficiency claims as approximate cross-paper comparisons.
minor comments (6)
  1. [Section I, Contributions list] There is a typo in 'We summarize the mian contributions as three-folds'; it should be 'main contributions'.
  2. [Section III.C, Eq. (1)] Eq. (1) has an unmatched parenthesis: F^{-1}(GELU(F(f^1 * W_{1x1}) * W_{1x1}) * W_{1x1} is missing a closing parenthesis for F^{-1}.
  3. [Section IV.D, Table V] The text says the 'Average' column summarizes five benchmarks (Set5, Set14, BSD100, Urban100, Manga109), but the table only shows BSD100, Urban100, and Manga109 individually; either include the missing columns or state how the average was computed.
  4. [Captions of Figs. 5-11 and text around Table II] Several figure captions say 'Quantitative evaluation results' when the figures show qualitative visual comparisons (e.g., Figs. 5, 6, 7, 8, 10, 11); conversely, the sentence preceding Table II says 'qualitative results' for a quantitative table. Please use 'qualitative' and 'quantitative' consistently.
  5. [References] The bibliography contains duplicate entries for the same work with different reference numbers: Restormer appears as [9] and [21], Uformer as [11] and [19], Rain200L as [12] and [41], DDN as [14] and [30], IDT as [18] and [104], and DRSformer as [10] and [38]. Please consolidate these into single references.
  6. [Section IV.A, Implementation Details] The sentence 'we utilize the outputs of the last three stages as nodes to categorize PW-FNet into small, medium and large scales' is vague; it is not clear which blocks or branches are retained or removed when switching between the S/M/L variants, nor how Table I derives distinct parameter counts of 0.719M, 1.196M, and 1.442M from this scheme.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: benchmark comparisons are external and the architecture is not derived from its own outputs; only minor same-group citations present.

full rationale

The paper's central claims are empirical: restoration quality and efficiency are tested on independent benchmarks (Rain200L/H, GoPro, Urban100, etc.) against external methods (Restormer, NAFNet, MambaIR, MaIR, etc.) in Tables I, IV, V, VI, and VII, so the quality comparison does not reduce to the paper's own definitions. The motivating pyramid Wavelet-Fourier pipeline (Section I, Fig. 2) is a hand-designed observation about where rain degradation concentrates; it informs the architecture, but no equation or fitted parameter forces the benchmark scores, and the architecture's components (PW-MIMO and the Fourier token mixer, Eqs. 1-2) are evaluated by ablations in Tables X-XIV. The Fourier-domain loss (Eq. 3) is a training objective, not a predicted quantity derived from itself. The same-group citations ([27] SFHformer and [40] FADformer) are prior works by overlapping authors, but they are used only as related work and as comparison baselines; the FADformer numbers in Tables I and XIV are externally published results, and no load-bearing uniqueness theorem or ansatz is imported from these citations. The internal inconsistency in FLOPs scaling (e.g., PW-FNet-S at 16.64G for 256x256 in Table I versus 55.6G for 512x512 in Table IX) and the unspecified FLOPs-counting methodology are correctness and robustness concerns, not circularity within the meaning of this review. Score 2 reflects the presence of minor same-group self-citations that are not load-bearing; no circular step was identified.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The paper introduces no new physical entities. Its load-bearing elements are hand-selected architecture choices, the motivating but unquantified wavelet-Fourier observation, and the comparability of external baseline numbers.

free parameters (5)
  • Pyramid levels and branch count = 3 levels, branches i=1,2,4
    Hand-chosen architecture configuration defining the multi-scale inputs and outputs.
  • Wavelet family = Daubechies
    Selected from ablation Table XI; Biorthogonal scored higher (42.43 vs 42.23 dB on Rain200L) but Daubechies was chosen for balance.
  • Fourier-domain loss with equal branch weights = L1 on complex FFT coefficients, weights 1 for branches 1,2,4
    Chosen via ablation Table XIII; Fourier loss alone beat all combinations with spatial or wavelet losses.
  • Training hyperparameters = 500K iterations, batch 24, patch 256, AdamW lr 1e-3 to 1e-6 cosine
    Standard recipe stated in Section IV.A; not tuned per task.
  • Network width and block count
    Not reported numerically; the parameter counts for S/M/L variants cannot be reconstructed from the text alone, which affects exact reproducibility.
assumptions (3)
  • domain assumption Degradation concentrates in high-frequency regions of the LL and HL wavelet-Fourier sub-bands.
    Motivating observation from Section I and Fig.2; presented as the basis for the design but not quantitatively established in the text.
  • domain assumption Global pointwise filtering in the Fourier domain can replace self-attention without losing long-range modeling.
    Architectural premise for the intra-block token mixer; only indirectly tested via the Fourier kernel size ablation in Table XII.
  • domain assumption Quoted baseline results from prior papers are directly comparable to the authors' own measurements.
    All quality and efficiency conclusions rely on this comparability, as stated in Section IV tables; baselines were not re-run.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Global Modeling Matters: A Fast, Lightweight and Effective Baseline for Efficient Image Restoration." pith.science (2026). https://pith.science/paper/R3ESOQSE

@misc{pith2026250713663,
  author       = {Pith},
  title        = {Pith review of: Global Modeling Matters: A Fast, Lightweight and Effective Baseline for Efficient Image Restoration},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/R3ESOQSE}},
  note         = {Machine review of arXiv:2507.13663}
}
read the original abstract

Natural image quality is often degraded by adverse weather conditions, significantly impairing the performance of downstream tasks. Image restoration has emerged as a core solution to this challenge and has been widely discussed in the literature. Although recent transformer-based approaches have made remarkable progress in image restoration, their increasing system complexity poses significant challenges for real-time processing, particularly in real-world deployment scenarios. To this end, most existing methods attempt to simplify the self-attention mechanism, such as by channel self-attention or state space model. However, these methods primarily focus on network architecture while neglecting the inherent characteristics of image restoration itself. In this context, we explore a pyramid Wavelet-Fourier iterative pipeline to demonstrate the potential of Wavelet-Fourier processing for image restoration. Inspired by the above findings, we propose a novel and efficient restoration baseline, named Pyramid Wavelet-Fourier Network (PW-FNet). Specifically, PW-FNet features two key design principles: 1) at the inter-block level, integrates a pyramid wavelet-based multi-input multi-output structure to achieve multi-scale and multi-frequency bands decomposition; and 2) at the intra-block level, incorporates Fourier transforms as an efficient alternative to self-attention mechanisms, effectively reducing computational complexity while preserving global modeling capability. Extensive experiments on tasks such as image deraining, raindrop removal, image super-resolution, motion deblurring, image dehazing, image desnowing and underwater/low-light enhancement demonstrate that PW-FNet not only surpasses state-of-the-art methods in restoration quality but also achieves superior efficiency, with significantly reduced parameter size, computational cost and inference time.

Figures

Figures reproduced from arXiv: 2507.13663 by the authors.

Figure 1
Figure 1. Comparison with different state-of-the-art approaches [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The pyramid wavelet-Fourier iterative pipeline for image restoration. Rain degradation is concentrated on HL sub-bands. [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Comparison of intra-block architecture designs for global modeling across various restoration baselines. [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Overall framework of our PW-FNet, which integrates wavelet and Fourier transforms into inter-block and intra-block. [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Quantitative evaluation results on the Rain200H [ [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Quantitative evaluation results on the real-world SPA-Data [ [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: Quantitative evaluation results on the Raindrop [ [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Quantitative evaluation results on the high-resolution 4K-Rain13K [ [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 9
Figure 9. Figure 9: Quantitative evaluation results on the super-resolution Urban100 [ [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 10
Figure 10. Figure 10: Quantitative evaluation results on the motion deblurring GoPro [ [PITH_FULL_IMAGE:figures/full_fig_p008_10.png]
Figure 11
Figure 11. Figure 11: Quantitative results on the downstream detection Rainy in Driving [ [PITH_FULL_IMAGE:figures/full_fig_p010_11.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

112 extracted references · 74 canonical work pages

  1. [1]

    Image-adaptive yolo for object detection in adverse weather conditions,

    W. Liu, G. Ren, R. Yu, S. Guo, J. Zhu, and L. Zhang, “Image-adaptive yolo for object detection in adverse weather conditions,” inProceedings of the AAAI conference on artificial intelligence , vol. 36, no. 2, 2022, pp. 1792–1800. 1

    2022

  2. [2]

    Adapnet: Adaptive semantic segmentation in adverse environmental conditions,

    A. Valada, J. Vertens, A. Dhall, and W. Burgard, “Adapnet: Adaptive semantic segmentation in adverse environmental conditions,” in 2017 IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 2017, pp. 4644–4651. 1

    2017

  3. [3]

    Vehicle detection and tracking in adverse weather using a deep learning framework,

    M. Hassaballah, M. A. Kenk, K. Muhammad, and S. Minaee, “Vehicle detection and tracking in adverse weather using a deep learning framework,” IEEE transactions on intelligent transportation systems , vol. 22, no. 7, pp. 4230–4242, 2020. 1

    2020

  4. [4]

    A generalized low-rank appearance model for spatio-temporally correlated rain streaks,

    Y .-L. Chen and C.-T. Hsu, “A generalized low-rank appearance model for spatio-temporally correlated rain streaks,” in Proceedings of the IEEE international conference on computer vision , 2013, pp. 1968–

    2013

  5. [5]

    Removing rain from a single image via discriminative sparse coding,

    Y . Luo, Y . Xu, and H. Ji, “Removing rain from a single image via discriminative sparse coding,” in Proceedings of the IEEE international conference on computer vision , 2015, pp. 3397–3405. 1, 7

    2015

  6. [6]

    Sapnet: Segmentation- aware progressive network for perceptual contrastive deraining,

    S. Zheng, C. Lu, Y . Wu, and G. Gupta, “Sapnet: Segmentation- aware progressive network for perceptual contrastive deraining,” in Proceedings of the IEEE/CVF winter conference on applications of computer vision, 2022, pp. 52–62. 1

    2022

  7. [7]

    Recurrent squeeze-and- excitation context aggregation net for single image deraining,

    X. Li, J. Wu, Z. Lin, H. Liu, and H. Zha, “Recurrent squeeze-and- excitation context aggregation net for single image deraining,” in Proceedings of the European conference on computer vision (ECCV) , 2018, pp. 254–269. 1

    2018

  8. [8]

    Progressive image deraining networks: A better and simpler baseline,

    D. Ren, W. Zuo, Q. Hu, P. Zhu, and D. Meng, “Progressive image deraining networks: A better and simpler baseline,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2019, pp. 3937–3946. 1

    2019

Show all 112 references
  1. [9]

    Restormer: Efficient transformer for high-resolution image restoration,

    S. W. Zamir, A. Arora, S. Khan, M. Hayat, F. S. Khan, and M.- H. Yang, “Restormer: Efficient transformer for high-resolution image restoration,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2022, pp. 5728–5739. 1, 4, 7, 9

    2022

  2. [10]

    Learning a sparse transformer net- work for effective image deraining,

    X. Chen, H. Li, M. Li, and J. Pan, “Learning a sparse transformer net- work for effective image deraining,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 5896–5905. 1, 9, 10

    2023

  3. [11]

    Uformer: A general u-shaped transformer for image restoration,

    Z. Wang, X. Cun, J. Bao, W. Zhou, J. Liu, and H. Li, “Uformer: A general u-shaped transformer for image restoration,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2022, pp. 17 683–17 693. 1, 3, 4, 7, 9

    2022

  4. [12]

    Deep joint rain detection and removal from a single image,

    W. Yang, R. T. Tan, J. Feng, J. Liu, Z. Guo, and S. Yan, “Deep joint rain detection and removal from a single image,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2017, pp. 1357–1366. 1, 5, 9

    2017

  5. [13]

    Multi- weather city: Adverse weather stacking for autonomous driving,

    V . Mus,at, I. Fursa, P. Newman, F. Cuzzolin, and A. Bradley, “Multi- weather city: Adverse weather stacking for autonomous driving,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 2906–2915. 1

    2021

  6. [14]

    Removing rain from single images via a deep detail network,

    X. Fu, J. Huang, D. Zeng, Y . Huang, X. Ding, and J. Paisley, “Removing rain from single images via a deep detail network,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 3855–3863. 3, 5

    2017

  7. [15]

    Recurrent squeeze-and- excitation context aggregation net for single image deraining,

    X. Li, J. Wu, Z. Lin, H. Liu, and H. Zha, “Recurrent squeeze-and- excitation context aggregation net for single image deraining,” in Proceedings of the European conference on computer vision (ECCV) , 2018, pp. 254–269. 3, 5

    2018

  8. [16]

    Progressive image deraining networks: A better and simpler baseline,

    D. Ren, W. Zuo, Q. Hu, P. Zhu, and D. Meng, “Progressive image deraining networks: A better and simpler baseline,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2019, pp. 3937–3946. 3, 5

    2019

  9. [17]

    Multi-scale progressive fusion network for single image deraining,

    K. Jiang, Z. Wang, P. Yi, C. Chen, B. Huang, Y . Luo, J. Ma, and J. Jiang, “Multi-scale progressive fusion network for single image deraining,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2020, pp. 8346–8355. 3, 5

    2020

  10. [18]

    Image de-raining transformer,

    J. Xiao, X. Fu, A. Liu, F. Wu, and Z.-J. Zha, “Image de-raining transformer,” IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022. 3, 5, 6, 7

    2022

  11. [19]

    Uformer: A general u-shaped transformer for image restoration,

    Z. Wang, X. Cun, J. Bao, W. Zhou, J. Liu, and H. Li, “Uformer: A general u-shaped transformer for image restoration,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2022, pp. 17 683–17 693. 3, 5

    2022

  12. [20]

    Swin transformer: Hierarchical vision transformer using shifted win- dows,

    Z. Liu, Y . Lin, Y . Cao, H. Hu, Y . Wei, Z. Zhang, S. Lin, and B. Guo, “Swin transformer: Hierarchical vision transformer using shifted win- dows,” in Proceedings of the IEEE/CVF international conference on computer vision, 2021, pp. 10 012–10 022. 3

    2021

  13. [21]

    Restormer: Efficient transformer for high-resolution image restoration,

    S. W. Zamir, A. Arora, S. Khan, M. Hayat, F. S. Khan, and M.- H. Yang, “Restormer: Efficient transformer for high-resolution image restoration,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2022, pp. 5728–5739. 3, 5, 6, 7

    2022

  14. [22]

    Metaformer is actually what you need for vision,

    W. Yu, M. Luo, P. Zhou, C. Si, Y . Zhou, X. Wang, J. Feng, and S. Yan, “Metaformer is actually what you need for vision,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2022, pp. 10 819–10 829. 3

    2022

  15. [23]

    Simple baselines for image restoration,

    L. Chen, X. Chu, X. Zhang, and J. Sun, “Simple baselines for image restoration,” in European Conference on Computer Vision . Springer, 2022, pp. 17–33. 3, 4, 7, 9

    2022

  16. [24]

    Dawn: Direction-aware attention wavelet network for image deraining,

    K. Jiang, W. Liu, Z. Wang, X. Zhong, J. Jiang, and C.-W. Lin, “Dawn: Direction-aware attention wavelet network for image deraining,” in Proceedings of the 31st ACM international conference on multimedia , 2023, pp. 7065–7074. 3, 10

    2023

  17. [25]

    Wave-mamba: Wavelet state space model for ultra-high-definition low-light image enhancement,

    W. Zou, H. Gao, W. Yang, and T. Liu, “Wave-mamba: Wavelet state space model for ultra-high-definition low-light image enhancement,” in Proceedings of the 32nd ACM International Conference on Multimedia, 2024, pp. 1534–1543. 3, 10

    2024

  18. [26]

    Selective frequency network for image restoration,

    Y . Cui, Y . Tao, Z. Bing, W. Ren, X. Gao, X. Cao, K. Huang, and A. Knoll, “Selective frequency network for image restoration,” in The Eleventh International Conference on Learning Representations , 2022. 3, 4, 7, 9

    2022

  19. [27]

    When fast fourier transform meets transformer for image restoration,

    X. Jiang, X. Zhang, N. Gao, and Y . Deng, “When fast fourier transform meets transformer for image restoration,” in European Conference on Computer Vision. Springer, 2024, pp. 381–402. 3

    2024

  20. [28]

    Learning dual- domain multi-scale representations for single image deraining,

    S. Zou, Y . Zou, M. Zhang, S. Luo, G. Gao, and G. Qi, “Learning dual- domain multi-scale representations for single image deraining,” arXiv preprint arXiv:2503.12014, 2025. 3, 10

    2025 arXiv

  21. [29]

    Focal network for image restoration,

    Y . Cui, W. Ren, X. Cao, and A. Knoll, “Focal network for image restoration,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2023, pp. 13 001–13 011. 4, 9

    2023

  22. [30]

    Removing rain from single images via a deep detail network,

    X. Fu, J. Huang, D. Zeng, Y . Huang, X. Ding, and J. Paisley, “Removing rain from single images via a deep detail network,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 3855–3863. 5

    2017

  23. [32]

    Spatial attentive single-image deraining with a high quality real rain dataset,

    T. Wang, X. Yang, K. Xu, S. Chen, Q. Zhang, and R. W. Lau, “Spatial attentive single-image deraining with a high quality real rain dataset,” JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 12 in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Re...

    2021

  24. [33]

    A model-driven deep neural network for single image rain removal,

    H. Wang, Q. Xie, Q. Zhao, and D. Meng, “A model-driven deep neural network for single image rain removal,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2020, pp. 3103–3112. 5, 7

    2020

  25. [34]

    Multi-stage progressive image restoration,

    S. W. Zamir, A. Arora, S. Khan, M. Hayat, F. S. Khan, M.-H. Yang, and L. Shao, “Multi-stage progressive image restoration,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2021, pp. 14 821–14 831. 5

    2021

  26. [35]

    Rain streak removal via dual graph convolutional network,

    X. Fu, Q. Qi, Z.-J. Zha, Y . Zhu, and X. Ding, “Rain streak removal via dual graph convolutional network,” in Proceedings of the AAAI Conference on Artificial Intelligence , vol. 35, no. 2, 2021, pp. 1352–

    2021

  27. [36]

    Structure- preserving deraining with residue channel prior guidance,

    Q. Yi, J. Li, Q. Dai, F. Fang, G. Zhang, and T. Zeng, “Structure- preserving deraining with residue channel prior guidance,” in Proceed- ings of the IEEE/CVF International Conference on Computer Vision , 2021, pp. 4238–4247. 5, 7

    2021

  28. [37]

    Hybrid cnn-transformer feature fusion for single image deraining,

    X. Chen, J. Pan, J. Lu, Z. Fan, and H. Li, “Hybrid cnn-transformer feature fusion for single image deraining,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 37, no. 1, 2023, pp. 378–386. 5

    2023

  29. [38]

    Learning a sparse transformer net- work for effective image deraining,

    X. Chen, H. Li, M. Li, and J. Pan, “Learning a sparse transformer net- work for effective image deraining,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 5896–5905. 5, 6, 7

    2023

  30. [39]

    Bidirectional multi-scale implicit neural representations for image deraining,

    X. Chen, J. Pan, and J. Dong, “Bidirectional multi-scale implicit neural representations for image deraining,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 25 627–25 636. 5, 6, 9, 10

    2024

  31. [40]

    Efficient frequency- domain image deraining with contrastive regularization,

    N. Gao, X. Jiang, X. Zhang, and Y . Deng, “Efficient frequency- domain image deraining with contrastive regularization,” in European Conference on Computer Vision . Springer, 2024, pp. 240–257. 5, 9, 10

    2024

  32. [41]

    Deep joint rain detection and removal from a single image,

    W. Yang, R. T. Tan, J. Feng, J. Liu, Z. Guo, and S. Yan, “Deep joint rain detection and removal from a single image,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2017, pp. 1357–1366. 5, 6, 10

    2017

  33. [42]

    Density-aware single image de-raining using a multi-stream dense network,

    H. Zhang and V . M. Patel, “Density-aware single image de-raining using a multi-stream dense network,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 695–

    2018

  34. [43]

    Deep learning for seeing through window with raindrops,

    Y . Quan, S. Deng, Y . Chen, and H. Ji, “Deep learning for seeing through window with raindrops,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2019, pp. 2463–2471. 5, 6

    2019

  35. [44]

    Spatial attentive single-image deraining with a high quality real rain dataset,

    T. Wang, X. Yang, K. Xu, S. Chen, Q. Zhang, and R. W. Lau, “Spatial attentive single-image deraining with a high quality real rain dataset,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 12 270–12 279. 5, 6

    2019

  36. [45]

    Towards ultra-high-definition image deraining: A benchmark and an efficient method,

    H. Chen, X. Chen, C. Wu, Z. Zheng, J. Pan, and X. Fu, “Towards ultra-high-definition image deraining: A benchmark and an efficient method,” arXiv preprint arXiv:2405.17074 , 2024. 5, 6, 7

    2024 arXiv

  37. [46]

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

    E. Agustsson and R. Timofte, “Ntire 2017 challenge on single image super-resolution: Dataset and study,” in Proceedings of the IEEE conference on computer vision and pattern recognition workshops , 2017, pp. 126–135. 5

    2017

  38. [47]

    Low-complexity single-image super-resolution based on nonnegative neighbor embedding,

    M. Bevilacqua, A. Roumy, C. Guillemot, and M. L. Alberi-Morel, “Low-complexity single-image super-resolution based on nonnegative neighbor embedding,” 2012. 5, 8

    2012

  39. [48]

    On single image scale-up using sparse-representations,

    R. Zeyde, M. Elad, and M. Protter, “On single image scale-up using sparse-representations,” in Curves and Surfaces: 7th International Con- ference, Avignon, France, June 24-30, 2010, Revised Selected Papers

    2010

  40. [49]

    Springer, 2012, pp. 711–730. 5, 8

    2012

  41. [50]

    A database of human seg- mented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics,

    D. Martin, C. Fowlkes, D. Tal, and J. Malik, “A database of human seg- mented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics,” in Proceedings eighth IEEE international conference on computer vision. ICCV 2001 , vol....

    2001

  42. [51]

    Sketch-based manga retrieval using manga109 dataset,

    Y . Matsui, K. Ito, Y . Aramaki, A. Fujimoto, T. Ogawa, T. Yamasaki, and K. Aizawa, “Sketch-based manga retrieval using manga109 dataset,” Multimedia tools and applications , vol. 76, pp. 21 811–21 838, 2017. 5, 8

    2017

  43. [52]

    Single image super-resolution from transformed self-exemplars,

    J.-B. Huang, A. Singh, and N. Ahuja, “Single image super-resolution from transformed self-exemplars,” in Proceedings of the IEEE confer- ence on computer vision and pattern recognition, 2015, pp. 5197–5206. 5, 8, 9

    2015

  44. [53]

    Deep multi-scale convolutional neural network for dynamic scene deblurring,

    S. Nah, T. Hyun Kim, and K. Mu Lee, “Deep multi-scale convolutional neural network for dynamic scene deblurring,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2017, pp. 3883–3891. 5, 7, 8, 9

    2017

  45. [54]

    Nh-haze: An image dehazing benchmark with non-homogeneous hazy and haze-free images,

    C. O. Ancuti, C. Ancuti, and R. Timofte, “Nh-haze: An image dehazing benchmark with non-homogeneous hazy and haze-free images,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops , 2020, pp. 444–445. 5, 9

    2020

  46. [55]

    Dense-haze: A benchmark for image dehazing with dense-haze and haze-free images,

    C. O. Ancuti, C. Ancuti, M. Sbert, and R. Timofte, “Dense-haze: A benchmark for image dehazing with dense-haze and haze-free images,” in 2019 IEEE international conference on image processing (ICIP) . IEEE, 2019, pp. 1014–1018. 5, 8, 9

    2019

  47. [56]

    Jstasr: Joint size and transparency-aware snow removal algorithm based on modified partial convolution and veiling effect removal,

    W.-T. Chen, H.-Y . Fang, J.-J. Ding, C.-C. Tsai, and S.-Y . Kuo, “Jstasr: Joint size and transparency-aware snow removal algorithm based on modified partial convolution and veiling effect removal,” in Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–...

    2020

  48. [57]

    Desnownet: Context-aware deep network for snow removal,

    Y .-F. Liu, D.-W. Jaw, S.-C. Huang, and J.-N. Hwang, “Desnownet: Context-aware deep network for snow removal,” IEEE Transactions on Image Processing , vol. 27, no. 6, pp. 3064–3073, 2018. 5, 8, 9

    2018

  49. [58]

    U-shape transformer for underwater image enhancement,

    L. Peng, C. Zhu, and L. Bian, “U-shape transformer for underwater image enhancement,” IEEE Transactions on Image Processing , 2023. 5, 8, 9

    2023

  50. [59]

    Sparse gradient regularized deep retinex network for robust low-light image enhance- ment,

    W. Yang, W. Wang, H. Huang, S. Wang, and J. Liu, “Sparse gradient regularized deep retinex network for robust low-light image enhance- ment,” IEEE Transactions on Image Processing , vol. 30, pp. 2072– 2086, 2021. 5, 9

    2021

  51. [60]

    Pytorch: An imperative style, high-performance deep learning library,

    A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga et al. , “Pytorch: An imperative style, high-performance deep learning library,” Advances in neural information processing systems , vol. 32, 2019. 5

    2019

  52. [61]

    Fixing weight decay regularization in adam,

    I. Loshchilov and F. Hutter, “Fixing weight decay regularization in adam,” 2018. 5

    2018

  53. [62]

    Sgdr: Stochastic gradient descent with warm restarts,

    ——, “Sgdr: Stochastic gradient descent with warm restarts,” arXiv preprint arXiv:1608.03983, 2016. 5

    2016 arXiv

  54. [63]

    Attentive generative adversarial network for raindrop removal from a single image,

    R. Qian, R. T. Tan, W. Yang, J. Su, and J. Liu, “Attentive generative adversarial network for raindrop removal from a single image,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 2482–2491. 6, 7

    2018

  55. [64]

    Image-to-image translation with conditional adversarial networks,

    P. Isola, J.-Y . Zhu, T. Zhou, and A. A. Efros, “Image-to-image translation with conditional adversarial networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 1125–1134. 6

    2017

  56. [65]

    Dual residual networks leveraging the potential of paired operations for image restoration,

    X. Liu, M. Suganuma, Z. Sun, and T. Okatani, “Dual residual networks leveraging the potential of paired operations for image restoration,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 7007–7016. 6

    2019

  57. [66]

    Maxim: Multi-axis mlp for image processing,

    Z. Tu, H. Talebi, H. Zhang, F. Yang, P. Milanfar, A. Bovik, and Y . Li, “Maxim: Multi-axis mlp for image processing,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 5769–5780. 6

    2022

  58. [67]

    Restoring vision in adverse weather conditions with patch-based denoising diffusion models,

    O. ¨Ozdenizci and R. Legenstein, “Restoring vision in adverse weather conditions with patch-based denoising diffusion models,” IEEE Trans- actions on Pattern Analysis and Machine Intelligence , 2023. 6

    2023

  59. [68]

    Sparse sampling transformer with uncertainty-driven ranking for unified removal of rain- drops and rain streaks,

    S. Chen, T. Ye, J. Bai, E. Chen, J. Shi, and L. Zhu, “Sparse sampling transformer with uncertainty-driven ranking for unified removal of rain- drops and rain streaks,” in Proceedings of the IEEE/CVF international conference on computer vision , 2023, pp. 13 106–13 117. 6, 7

    2023

  60. [69]

    Adapt or perish: Adaptive sparse transformer with attentive feature refinement for image restoration,

    S. Zhou, D. Chen, J. Pan, J. Shi, and J. Yang, “Adapt or perish: Adaptive sparse transformer with attentive feature refinement for image restoration,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 2952–2963. 6

    2024

  61. [70]

    Lightweight pyramid networks for image deraining,

    X. Fu, B. Liang, Y . Huang, X. Ding, and J. Paisley, “Lightweight pyramid networks for image deraining,” IEEE transactions on neural networks and learning systems , vol. 31, no. 6, pp. 1794–1807, 2019. 7

    2019

  62. [71]

    Joint rain detection and removal from a single image with contextualized deep networks,

    W. Yang, R. T. Tan, J. Feng, Z. Guo, S. Yan, and J. Liu, “Joint rain detection and removal from a single image with contextualized deep networks,” IEEE transactions on pattern analysis and machine intelligence, vol. 42, no. 6, pp. 1377–1393, 2019. 7

    2019

  63. [72]

    Rethinking coarse-to-fine approach in single image deblurring,

    S.-J. Cho, S.-W. Ji, J.-P. Hong, S.-W. Jung, and S.-J. Ko, “Rethinking coarse-to-fine approach in single image deblurring,” in Proceedings of the IEEE/CVF international conference on computer vision , 2021, pp. 4641–4650. 7

    2021

  64. [73]

    Multi-stage progressive image restoration,

    S. W. Zamir, A. Arora, S. Khan, M. Hayat, F. S. Khan, M.-H. Yang, and L. Shao, “Multi-stage progressive image restoration,” in Proceedings of JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 13 the IEEE/CVF conference on computer vision and pattern recognition , 2021,...

    2021

  65. [74]

    A comparative study of image restoration networks for general back- bone network design,

    X. Chen, Z. Li, Y . Pu, Y . Liu, J. Zhou, Y . Qiao, and C. Dong, “A comparative study of image restoration networks for general back- bone network design,” in European Conference on Computer Vision . Springer, 2024, pp. 74–91. 7, 9

    2024

  66. [75]

    Acl: Activating capability of linear attention for image restoration,

    Y . Gu, Y . Meng, J. Ji, and X. Sun, “Acl: Activating capability of linear attention for image restoration,” in Proceedings of the Computer Vision and Pattern Recognition Conference , 2025, pp. 17 913–17 923. 7

    2025

  67. [76]

    Mair: A locality-and continuity-preserving mamba for image restoration,

    B. Li, H. Zhao, W. Wang, P. Hu, Y . Gou, and X. Peng, “Mair: A locality-and continuity-preserving mamba for image restoration,” in Proceedings of the Computer Vision and Pattern Recognition Confer- ence, 2025, pp. 7491–7501. 7, 8, 9

    2025

  68. [77]

    Mambair: A simple baseline for image restoration with state-space model,

    H. Guo, J. Li, T. Dai, Z. Ouyang, X. Ren, and S.-T. Xia, “Mambair: A simple baseline for image restoration with state-space model,” in European conference on computer vision . Springer, 2024, pp. 222–

    2024

  69. [78]

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

    N. Ahn, B. Kang, and K.-A. Sohn, “Fast, accurate, and lightweight super-resolution with cascading residual network,” in Proceedings of the European conference on computer vision (ECCV) , 2018, pp. 252–

    2018

  70. [79]

    Efficient long-range attention network for image super-resolution,

    X. Zhang, H. Zeng, S. Guo, and L. Zhang, “Efficient long-range attention network for image super-resolution,” in European conference on computer vision . Springer, 2022, pp. 649–667. 8, 9

    2022

  71. [80]

    Swinir: Image restoration using swin transformer,

    J. Liang, J. Cao, G. Sun, K. Zhang, L. Van Gool, and R. Timofte, “Swinir: Image restoration using swin transformer,” in Proceedings of the IEEE/CVF international conference on computer vision , 2021, pp. 1833–1844. 8, 9

    2021

  72. [81]

    Srformer: Permuted self-attention for single image super-resolution,

    Y . Zhou, Z. Li, C.-L. Guo, S. Bai, M.-M. Cheng, and Q. Hou, “Srformer: Permuted self-attention for single image super-resolution,” in Proceedings of the IEEE/CVF international conference on computer vision, 2023, pp. 12 780–12 791. 8, 9

    2023

  73. [82]

    Multi-scale boosted dehazing network with dense feature fusion,

    H. Dong, J. Pan, L. Xiang, Z. Hu, X. Zhang, F. Wang, and M.-H. Yang, “Multi-scale boosted dehazing network with dense feature fusion,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 2157–2167. 9

    2020

  74. [83]

    Pre-trained image processing transformer,

    H. Chen, Y . Wang, T. Guo, C. Xu, Y . Deng, Z. Liu, S. Ma, C. Xu, C. Xu, and W. Gao, “Pre-trained image processing transformer,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 12 299–12 310. 9

    2021

  75. [84]

    A wavelet-based dual-stream network for under- water image enhancement,

    Z. Ma and C. Oh, “A wavelet-based dual-stream network for under- water image enhancement,” in ICASSP 2022-2022 IEEE international conference on acoustics, speech and signal processing (ICASSP) . IEEE, 2022, pp. 2769–2773. 9

    2022

  76. [85]

    Ffa-net: Feature fusion attention network for single image dehazing,

    X. Qin, Z. Wang, Y . Bai, X. Xie, and H. Jia, “Ffa-net: Feature fusion attention network for single image dehazing,” in Proceedings of the AAAI conference on artificial intelligence , vol. 34, no. 07, 2020, pp. 11 908–11 915. 9

    2020

  77. [86]

    Cycle-dehaze: Enhanced cyclegan for single image dehazing,

    D. Engin, A. Genc ¸, and H. Kemal Ekenel, “Cycle-dehaze: Enhanced cyclegan for single image dehazing,” in Proceedings of the IEEE conference on computer vision and pattern recognition workshops , 2018, pp. 825–833. 9

    2018

  78. [87]

    Retinex-inspired unrolling with cooperative prior architecture search for low-light im- age enhancement,

    R. Liu, L. Ma, J. Zhang, X. Fan, and Z. Luo, “Retinex-inspired unrolling with cooperative prior architecture search for low-light im- age enhancement,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 10 561–10 570. 9

    2021

  79. [88]

    Deep underwater image enhancement,

    S. Anwar, C. Li, and F. Porikli, “Deep underwater image enhancement,” arXiv preprint arXiv:1807.03528 , 2018. 9

    2018 arXiv

  80. [89]

    Image dehazing transformer with transmission-aware 3d position embedding,

    C.-L. Guo, Q. Yan, S. Anwar, R. Cong, W. Ren, and C. Li, “Image dehazing transformer with transmission-aware 3d position embedding,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 5812–5820. 9

    2022

  81. [90]

    Perceiving and modeling density for image dehazing,

    T. Ye, Y . Zhang, M. Jiang, L. Chen, Y . Liu, S. Chen, and E. Chen, “Perceiving and modeling density for image dehazing,” in European Conference on Computer Vision . Springer, 2022, pp. 130–145. 9

    2022

  82. [91]

    All snow removed: Single image desnowing algo- rithm using hierarchical dual-tree complex wavelet representation and contradict channel loss,

    W.-T. Chen, H.-Y . Fang, C.-L. Hsieh, C.-C. Tsai, I. Chen, J.-J. Ding, S.-Y . Kuo et al. , “All snow removed: Single image desnowing algo- rithm using hierarchical dual-tree complex wavelet representation and contradict channel loss,” in Proceedings of the IEEE/CVF Internation...

    2021

  83. [92]

    Band representation- based semi-supervised low-light image enhancement: Bridging the gap between signal fidelity and perceptual quality,

    W. Yang, S. Wang, Y . Fang, Y . Wang, and J. Liu, “Band representation- based semi-supervised low-light image enhancement: Bridging the gap between signal fidelity and perceptual quality,” IEEE Transactions on Image Processing, vol. 30, pp. 3461–3473, 2021. 9

    2021

  84. [93]

    Underwater image enhancement via learning water type desensitized representations,

    Z. Fu, X. Lin, W. Wang, Y . Huang, and X. Ding, “Underwater image enhancement via learning water type desensitized representations,” in ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2022, pp. 2764–2768. 9

    2022

  85. [94]

    Underwater image enhancement via medium transmission-guided multi-color space embedding,

    C. Li, S. Anwar, J. Hou, R. Cong, C. Guo, and W. Ren, “Underwater image enhancement via medium transmission-guided multi-color space embedding,” IEEE Transactions on Image Processing , vol. 30, pp. 4985–5000, 2021. 9

    2021

  86. [95]

    Curricular contrastive regularization for physics-aware single image dehazing,

    Y . Zheng, J. Zhan, S. He, J. Dong, and Y . Du, “Curricular contrastive regularization for physics-aware single image dehazing,” in Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 5785–5794. 9

    2023

  87. [96]

    Learning enriched features for real image restoration and enhancement,

    S. W. Zamir, A. Arora, S. Khan, M. Hayat, F. S. Khan, M.-H. Yang, and L. Shao, “Learning enriched features for real image restoration and enhancement,” in Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XXV

    2020

  88. [97]

    Springer, 2020, pp. 492–511. 9

    2020

  89. [98]

    An underwater image enhancement benchmark dataset and beyond,

    C. Li, C. Guo, W. Ren, R. Cong, J. Hou, S. Kwong, and D. Tao, “An underwater image enhancement benchmark dataset and beyond,” IEEE transactions on image processing , vol. 29, pp. 4376–4389, 2019. 9

    2019

  90. [99]

    Fourmer: an efficient global modeling paradigm for image restoration,

    M. Zhou, J. Huang, C.-L. Guo, and C. Li, “Fourmer: an efficient global modeling paradigm for image restoration,” in International Conference on Machine Learning . PMLR, 2023, pp. 42 589–42 601. 9

    2023

  91. [100]

    Irnext: Rethinking convolutional network design for image restoration,

    Y . Cui, W. Ren, S. Yang, X. Cao, and A. Knoll, “Irnext: Rethinking convolutional network design for image restoration,” 2023. 9

    2023

  92. [101]

    Snr-aware low-light image en- hancement,

    X. Xu, R. Wang, C.-W. Fu, and J. Jia, “Snr-aware low-light image en- hancement,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2022, pp. 17 714–17 724. 9

    2022

  93. [102]

    Underwater image enhance- ment by transformer-based diffusion model with non-uniform sampling for skip strategy,

    Y . Tang, H. Kawasaki, and T. Iwaguchi, “Underwater image enhance- ment by transformer-based diffusion model with non-uniform sampling for skip strategy,” in Proceedings of the 31st ACM international conference on multimedia , 2023, pp. 5419–5427. 9

    2023

  94. [103]

    Omni-kernel network for image restora- tion,

    Y . Cui, W. Ren, and A. Knoll, “Omni-kernel network for image restora- tion,” in Proceedings of the AAAI conference on artificial intelligence , vol. 38, no. 2, 2024, pp. 1426–1434. 9

    2024

  95. [104]

    Retinex- former: One-stage retinex-based transformer for low-light image en- hancement,

    Y . Cai, H. Bian, J. Lin, H. Wang, R. Timofte, and Y . Zhang, “Retinex- former: One-stage retinex-based transformer for low-light image en- hancement,” arXiv preprint arXiv:2303.06705 , 2023. 9

    2023 arXiv

  96. [105]

    Wavelet-based fourier infor- mation interaction with frequency diffusion adjustment for underwater image restoration,

    C. Zhao, W. Cai, C. Dong, and C. Hu, “Wavelet-based fourier infor- mation interaction with frequency diffusion adjustment for underwater image restoration,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 8281–8291. 9

    2024

  97. [106]

    Image de-raining transformer,

    J. Xiao, X. Fu, A. Liu, F. Wu, and Z.-J. Zha, “Image de-raining transformer,” IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022. 9, 10

    2022

  98. [107]

    Not just streaks: Towards ground truth for single image deraining,

    Y . Ba, H. Zhang, E. Yang, A. Suzuki, A. Pfahnl, C. C. Chandrappa, C. M. De Melo, S. You, S. Soatto, A. Wong et al. , “Not just streaks: Towards ground truth for single image deraining,” in European Conference on Computer Vision . Springer, 2022, pp. 723–740. 9

    2022

  99. [108]

    From sky to the ground: A large-scale benchmark and simple baseline towards real rain removal,

    Y . Guo, X. Xiao, Y . Chang, S. Deng, and L. Yan, “From sky to the ground: A large-scale benchmark and simple baseline towards real rain removal,” in Proceedings of the IEEE/CVF international conference on computer vision, 2023, pp. 12 097–12 107. 9

    2023

  100. [109]

    Making a “completely blind

    A. Mittal, R. Soundararajan, and A. C. Bovik, “Making a “completely blind” image quality analyzer,” IEEE Signal processing letters, vol. 20, no. 3, pp. 209–212, 2012. 9

    2012

  101. [110]

    No-reference image quality assessment in the spatial domain,

    A. Mittal, A. K. Moorthy, and A. C. Bovik, “No-reference image quality assessment in the spatial domain,” IEEE Transactions on image processing, vol. 21, no. 12, pp. 4695–4708, 2012. 9

    2012

  102. [111]

    Single image deraining: A comprehensive benchmark analysis,

    S. Li, I. B. Araujo, W. Ren, Z. Wang, E. K. Tokuda, R. H. Junior, R. Cesar-Junior, J. Zhang, X. Guo, and X. Cao, “Single image deraining: A comprehensive benchmark analysis,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2019, pp. 3838–3...

    2019

  103. [112]

    Yolov6: A single-stage object detection framework for industrial applications,

    C. Li, L. Li, H. Jiang, K. Weng, Y . Geng, L. Li, Z. Ke, Q. Li, M. Cheng, W. Nie et al., “Yolov6: A single-stage object detection framework for industrial applications,” arXiv preprint arXiv:2209.02976 , 2022. 9

    2022 arXiv

  104. [113]

    Efficient frequency domain-based transformers for high-quality image deblurring,

    L. Kong, J. Dong, J. Ge, M. Li, and J. Pan, “Efficient frequency domain-based transformers for high-quality image deblurring,” in Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 5886–5895. 10

    2023

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.