Pith. sign in

REVIEW 3 major objections 4 minor 2 cited by

Fast and Accurate Image Restoration and Generation with Rank Enhanced Linear Attention

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Rank-boosted linear attention wins 21 restoration benchmarks

desk verdict Solid empirical contribution with an oversold rank-enhancement story; worth reviewing, but the mechanism claim needs a 1×1 DWC ablation and a caveat. read the letter →

arxiv 2505.16157 v2 pith:H6TAQH3O submitted 2025-05-22 cs.CV

Yuang Ai , Huaibo Huang , Tao Wu , Qihang Fan , Ran He This is my paper · ORCID
classification cs.CV
keywords linearattentionimagerestorationlow-rankdepthwiseconvolutionefficienttransformerhigh-resolutionall-in-onemotiondeblurring
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 show that linear attention, long considered too weak for image restoration, can replace quadratic softmax attention without losing quality, as long as its output features are made full-rank again. The proposed fix, Rank Enhanced Linear Attention (RELA), adds a single lightweight depthwise convolution to the linear-attention computation, and the resulting LAformer is claimed to reach state-of-the-art results on seven restoration tasks and 21 benchmarks while keeping complexity linear in the number of pixels. On motion deblurring, LAformer-B reports 33.40 dB PSNR on GoPro, above Stripformer (33.08 dB) and Restormer (32.92 dB), at comparable or lower FLOPs and runtime. The same block is also claimed to work as a backbone for diffusion- and flow-based image generation, offering an efficient alternative to DiT and SiT. If these results hold, high-resolution restoration no longer needs window shifting or sparse attention to stay tractable.

What carries the argument

The central object is Rank Enhanced Linear Attention (RELA), defined by joining the linear attention output with a depthwise convolution of the value features: $Y = \phi(Q)\phi(K)^T V + W_d V$, with $\phi(\cdot) = 1+\mathrm{ELU}(\cdot)$. The depthwise convolution acts as a fixed local attention that mixes nearby pixels, breaking the rank ceiling that linear attention's global map imposes; the paper demonstrates the rank restoration empirically rather than by a formal rank proof for the sum. Around RELA, the Dual Attention Block combines it with a Channel Attention Block (global channel statistics) and a Convolutional Gated FFN (local detail fitting), all of which keep the architecture free of softmax and window-shifting operations so that cost stays linear in token count.

What would settle it

Run the released LAformer on the GoPro test split under the paper's protocol and compare PSNR against the reported 33.40 dB; then, on SOTS-Indoor, retrain with RELA's depthwise convolution replaced by a 1×1 convolution of equal parameter count. The paper's rank explanation is falsified if PSNR stays within 0.1 dB while the numerical rank of block outputs drops back to the low-rank linear-attention level; the empirical SOTA claim is falsified if a faithful re-run misses 33.40 dB by more than about 0.2 dB.

Watch

Extended reading notes

Core claim

The paper's central claim is that vanilla linear attention underperforms in image restoration because its attention map $\psi(Q)\psi(K)^T$ has rank at most $\min(N, C)$, and since high-resolution images have far more tokens $N$ than channels $C$, the output features collapse to a low-rank subspace. RELA repairs this by computing $Y = (1+\mathrm{ELU}(Q))\,(1+\mathrm{ELU}(K))^T V + W_d V$, where $W_d$ is a $5\times5$ depthwise convolution that acts as a local attention branch; the paper reports that this restores the numerical rank of output features to full rank across every block. With RELA and a channel-attention branch plus a convolutional gated feed-forward network, LAformer is claimed to surpass prior state-of-the-art on all evaluated tasks, including gains of 0.32 dB over Histoformer on raindrop removal and 0.87 dB over GRL on RealBlur-R, while avoiding softmax and window shifting entirely.

Load-bearing premise

The load-bearing premise is that the low-rank attention map is the main reason vanilla linear attention fails in restoration, and that one fixed depthwise convolution restores enough feature diversity to close the gap; the paper's own ablations suggest the convolution's exact form hardly matters, so the mechanism could be generic local feature extraction instead.

Editorial extensions

If this is right

  • High-resolution images (e.g., 1280×720 deblurring) can be processed with a global receptive field at linear complexity, with LAformer-B running at 0.76 s per GoPro frame versus 0.79 s for Restormer at similar parameter count.
  • A single LAformer trained all-in-one on five degradations beats the diffusion-based DiffUIR on all five, so task-specific attention design is not required for state-of-the-art restoration.
  • The RELA block is presented as a drop-in backbone for diffusion and flow generation, meaning the same linear-complexity design could scale to megapixel synthesis.
  • Removing softmax and window shifting eliminates two hardware-inefficient operations, making the architecture more GPU- and edge-friendly at high resolutions.

Reading between the lines

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

  • Editorial inference: the rank story is only loosely load-bearing. The paper's Table 12 shows that changing RELA's depthwise kernel from 3×3 to 7×7 shifts GoPro PSNR by only 0.06 dB and switching activations by 0.02 dB, so the convolution may be acting as a generic local feature extractor rather than a precise rank-restoration mechanism.
  • Editorial inference: if that is right, simpler or cheaper local operators than a 5×5 depthwise convolution could give the same gains; replacing $W_d$ with a 1×1 convolution or a fixed high-pass filter and measuring both output rank and PSNR would separate the two explanations.
  • Editorial inference: the complexity advantage grows with resolution, so the most decisive tests of the paper's thesis are the largest images; applying LAformer to 4K video frames or gigapixel restoration would show whether the linear-complexity claim translates to real speedups.
  • Editorial inference: the all-in-one results suggest the global-plus-local mix is degradation-agnostic; extending to unseen corruption combinations such as blur plus haze plus noise jointly would test whether the representation capacity holds beyond the five trained tasks.
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

3 major / 4 minor

Summary. The paper proposes Rank Enhanced Linear Attention (RELA), a linear-attention variant that adds a depthwise-convolution branch, and builds LAformer, a U-Net-style restoration transformer combining RELA with channel attention and a convolutional gated feed-forward network. The authors claim state-of-the-art results across 7 image restoration tasks and 21 benchmarks, report favorable efficiency compared with softmax/window transformers, and state in the abstract that the approach also extends to diffusion- and flow-based visual generation. The central conceptual claim is that vanilla linear attention fails in image restoration because its attention map is low rank, and that the depthwise convolution in RELA enriches output features by restoring their rank.

Significance. If the reported numbers are verified, LAformer is a practically useful architecture: it replaces softmax/window attention with linear attention plus local depthwise mixing, obtains strong results across many restoration benchmarks, and reports a favorable FLOP/latency trade-off. The paper's strengths are the breadth of evaluation, the component-level ablations in Tables 11 and 12, and the promise of public code. However, the central conceptual novelty—that depthwise convolution enhances the rank of linear-attention output features—is not tightly supported by the paper's own analysis, and the generation claim in the abstract is unsupported by the presented experiments. Both issues need to be resolved before the paper's framing matches its evidence.

major comments (3)
  1. [§3.2, Eq. (3), Fig. 4, Table 12] The rank-enhancement justification is not established by the presented evidence. Eq. (3) bounds the rank of the attention map M = ψ(Q)ψ(K)^T by C, but the output Y = M V is N×C and can have rank up to C by construction; the observed rank 24 vs. 48 in Fig. 4 is therefore an empirical property of learned features, not a consequence of the factorization alone. Table 12 amplifies this concern: changing the depthwise convolution kernel from 3×3 to 7×7 changes GoPro PSNR by only 0.06 dB, and changing the activation from ReLU to 1+ELU changes it by 0.02 dB. This is consistent with the depthwise convolution acting as a generic local feature extractor rather than as a mechanism specifically tied to output-feature rank. Please either provide direct evidence that the output features are rank-deficient and that the DWC specifically restores rank (e.g., report numerical rank with a defined threshold, and test whether a 1×1 DWC or per-channel scaling—which would also restore algebraic rank—recovers the benefit), or reframe RELA as an empirical design that combines linear attention with local mixing and remove the implication that Eq. (3) predicts the failure.
  2. [Abstract and §5 (Conclusion)] The abstract states that LAformer is extended to diffusion-based and flow-based visual generation and that it is a competitive alternative to DiT and SiT, but the manuscript contains no generation experiments, no architecture details for such an extension, and no results tables. Since this is a forward-looking claim in the abstract and conclusion, it is load-bearing for the paper's stated scope. Either add the generation experiments and the required implementation details, or remove the claim from the abstract and conclusion so that the paper only claims image-restoration results.
  3. [Table 6, Outdoor-Haze row] The reported SSIM of 0.932 for LAformer-T on Outdoor-Haze is implausible given the surrounding rows: PMNet reports 0.830 and MB-TaylorFormer 0.788, with PSNR values only 0.64 dB and 0.23 dB below LAformer-T's 25.28 dB. A PSNR gap of 0.23 dB cannot plausibly produce an SSIM jump of 0.144. This appears to be an error (possibly a swapped or miscomputed metric). Because the dehazing SOTA claim depends on this table, please verify or correct the entry and re-check all tables for similar anomalies.
minor comments (4)
  1. [Table 2, References] Restormer is cited as [93] in Table 2 but as [115] elsewhere, and reference [93] is actually PromptRestorer. Please correct the citation to [115].
  2. [Table 11] The ablation row 'w/o CAB' reports FLOPs of 30.79 G, which is higher than the full Dual Attention row at 29.84 G even though CAB is removed. This is counterintuitive and should be checked; if the row corresponds to a different replacement, the text should say so.
  3. [§4.2 and Tables 1–10] The paper claims '7 tasks and 21 benchmarks' but does not explicitly enumerate which datasets count toward the 21. A summary table or list would improve traceability of this claim.
  4. [Fig. 4 and Fig. 7] The rank measurements are not defined: no numerical threshold or singular-value criterion is given for computing 'rank' from feature maps. Please specify the procedure used to obtain the rank values.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: RELA is an architectural modification validated on held-out benchmarks; the rank analysis is motivational rather than a fitted prediction.

full rationale

The paper's central claim is an architectural proposal (RELA = linear attention plus a depthwise convolution) evaluated on held-out test sets across seven restoration tasks. No parameter is fitted to the benchmark targets and then relabeled as a prediction; the only tuned design choice, the 5x5 DWC kernel, is ablated on GoPro with a 0.05 dB spread (Tab. 12), so the final PSNR is not statistically forced by that choice. The rank argument in Sec. 3.2 is a post-hoc motivation: Eq. (3) bounds the attention map by min(N,C), and the observed output rank (24 vs 48 in Fig. 4) is empirical, not a consequence of the factorization alone. The rank-restoring effect of the DWC is measured, not assumed, and the performance gains are measured on independent datasets. The only self-citation ([4], MPerceiver) appears as a comparison baseline and is not load-bearing. The fact that the rank mechanism may not be tightly tied to performance (Tab. 12 shows kernel size and activation have small effects) is a correctness/interpretation concern, not a circularity concern.

Assumptions & free parameters 2 free parameters · 2 assumptions · 0 invented entities

The paper introduces no new theoretical entities. Its ledger consists of the rank-enhancement assumption and standard, small hyperparameters. The main dependence is on benchmark datasets and empirical tuning.

free parameters (2)
  • Depthwise convolution kernel size in RELA = 5x5
    Selected based on the ablation in Table 12, where 3x3, 5x5, and 7x7 give PSNR 33.35, 33.40, and 33.41 dB. The choice is a hand-tuned hyperparameter, but its effect is small.
  • Model width and depth variants (T/S/B) = Not specified in main text; varied per task
    The paper states that the number of blocks and channels are adapted per task to match SOTA models, but exact configurations are deferred to the appendix, which is missing.
assumptions (2)
  • domain assumption Rank of the linear attention map limits output feature diversity, and restoring full rank improves restoration quality.
    This is the core motivation of RELA, introduced in Sec. 3.2. The rank analysis in Fig. 7 shows a correlation, but the ablations in Tab. 12 do not show a strong dependence on the specific rank-enhancing mechanism, so the causal link is an assumption.
  • domain assumption Standard benchmark datasets and training protocols from prior work are valid for measuring restoration quality.
    The paper relies on established datasets (GoPro, SOTS, DPDD, etc.) and reports PSNR/SSIM, which are standard but known to be imperfect measures of perceptual quality.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Fast and Accurate Image Restoration and Generation with Rank Enhanced Linear Attention." pith.science (2026). https://pith.science/paper/H6TAQH3O

@misc{pith2026250516157,
  author       = {Pith},
  title        = {Pith review of: Fast and Accurate Image Restoration and Generation with Rank Enhanced Linear Attention},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/H6TAQH3O}},
  note         = {Machine review of arXiv:2505.16157}
}
read the original abstract

Transformer-based models have made remarkable progress in image restoration (IR) tasks. However, the quadratic complexity of self-attention in Transformer hinders its applicability to high-resolution images. Existing methods mitigate this issue with sparse or window-based attention, yet inherently limit global context modeling. Linear attention, a variant of softmax attention, demonstrates promise in global context modeling while maintaining linear complexity, offering a potential solution to the above challenge. Despite its efficiency benefits, vanilla linear attention suffers from a significant performance drop in IR, largely due to the low-rank nature of its attention map. To counter this, we propose Rank Enhanced Linear Attention (RELA), a simple yet effective method that enriches feature representations by integrating a lightweight depthwise convolution. Building upon RELA, we propose an efficient and effective Vision Transformer, named LAformer. LAformer eliminates hardware-inefficient operations such as softmax and window shifting, enabling efficient processing of high-resolution images. Extensive experiments across 7 IR tasks and 21 benchmarks demonstrate that LAformer outperforms SOTA methods and offers significant computational advantages. Furthermore, we extend LAformer to diffusion-based and flow-based visual generation, showcasing its strong potential as a competitive alternative to DiT and SiT. Code and models are available at https://github.com/shallowdream204/LAformer.

Figures

Figures reproduced from arXiv: 2505.16157 by the authors.

Figure 1
Figure 1. Comparison with Transformer-based methods for rain [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Conceptual comparison between Softmax Attention and Linear Attention. Instead of using softmax, linear attention employs simpler activation functions (e.g., ReLU) to decouple the process, altering the computation order to first calculate KT V . While linear attention excels at efficiently capturing global con￾text, its representation capacity is limited compared to softmax attention, resulting in a notable performan… view at source ↗
Figure 3
Figure 3. Overall architecture of the proposed LAformer, which includes (b) Dual Attention (DA) Block, (c) Rank Enhanced Linear [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Comparison of feature maps (N = 256, C = 48) output by softmax attention (window-based), linear attention, and RELA. The output features of linear attention often exhibit pronounced low-rank characteristics, which constrain its representational ca￾pacity. Our proposed …
Figure 5
Figure 5. Figure 5: Top row: Visual comparison on the DPDD [ [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Top row: Visual comparison on the LOL-v1 [ [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Rank analysis across model blocks for different attention [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Rectifying Magnitude Neglect in Linear Attention

    cs.CV 2025-07 conditional novelty 6.0 of 10

    MALA adds a scale-sensitive offset to linear attention so attention sharpens with query magnitude, improving accuracy across several tasks.

  2. A Unified Resolution-Conditioned Framework for Orthogonal Line-Scanning Image Fusion

    cs.CV 2026-08 conditional novelty 5.0 of 10

    A resolution-conditioned linear-attention network fuses orthogonal line scans across slit widths in one model, outperforming unconditioned multi-slit training and per-slit specialists on simulated data.

Reference graph

Works this paper leans on

131 extracted references · 70 canonical work pages · cited by 2 Pith papers

  1. [93]

    Promptre- storer: A prompting image restoration method with degra- dation perception

    Cong Wang, Jinshan Pan, Wei Wang, Jiangxin Dong, Mengzhu Wang, Yakun Ju, and Junyang Chen. Promptre- storer: A prompting image restoration method with degra- dation perception. InNeurIPS, pages 8898–8912, 2023. 5

    work page 2023

  2. [115]

    Restormer: Efficient transformer for high-resolution image restoration

    Syed Waqas Zamir, Aditya Arora, Salman Khan, Mu- nawar Hayat, Fahad Shahbaz Khan, and Ming-Hsuan Yang. Restormer: Efficient transformer for high-resolution image restoration. InCVPR, pages 5728–5739, 2022. 1, 2, 4, 5, 6, 7, 8

    work page 2022

  3. [1]

    Defocus deblur- ring using dual-pixel data

    Abdullah Abuolaim and Michael S Brown. Defocus deblur- ring using dual-pixel data. InECCV, pages 111–126, 2020. 6

    2020

  4. [2]

    Brown, and Peyman Milanfar

    Abdullah Abuolaim, Mauricio Delbracio, Damien Kelly, Michael S. Brown, and Peyman Milanfar. Learning to re- duce defocus blur by realistically modeling dual-pixel data. InICCV, 2021. 6

    2021

  5. [3]

    Deep learning using rectified linear units (relu)

    AF Agarap. Deep learning using rectified linear units (relu). arXiv preprint arXiv:1803.08375, 2018. 2

    arXiv 2018

  6. [4]

    Multimodal prompt perceiver: Empower adap- tiveness generalizability and fidelity for all-in-one image restoration

    Yuang Ai, Huaibo Huang, Xiaoqiang Zhou, Jiexiang Wang, and Ran He. Multimodal prompt perceiver: Empower adap- tiveness generalizability and fidelity for all-in-one image restoration. InCVPR, pages 25432–25444, 2024. 6

    2024

  7. [5]

    O-haze: a dehazing bench- mark with real hazy and haze-free outdoor images

    Codruta O Ancuti, Cosmin Ancuti, Radu Timofte, and Christophe De Vleeschouwer. O-haze: a dehazing bench- mark with real hazy and haze-free outdoor images. In CVPRW, pages 754–762, 2018. 6

    2018

  8. [6]

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

    Codruta O Ancuti, Cosmin Ancuti, Mateu Sbert, and Radu Timofte. Dense-haze: A benchmark for image dehazing with dense-haze and haze-free images. InICIP, pages 1014–1018, 2019. 6, 7

    2019

Show all 131 references
  1. [7]

    Hydra attention: Efficient at- tention with many heads

    Daniel Bolya, Cheng-Yang Fu, Xiaoliang Dai, Peizhao Zhang, and Judy Hoffman. Hydra attention: Efficient at- tention with many heads. InECCV, pages 35–49, 2022. 3

    2022

  2. [8]

    Dehazenet: An end-to-end system for sin- gle image haze removal.TIP, 25(11):5187–5198, 2016

    Bolun Cai, Xiangmin Xu, Kui Jia, Chunmei Qing, and Dacheng Tao. Dehazenet: An end-to-end system for sin- gle image haze removal.TIP, 25(11):5187–5198, 2016. 6

    2016

  3. [9]

    Efficientvit: Lightweight multi-scale attention for high-resolution dense prediction

    Han Cai, Junyan Li, Muyan Hu, Chuang Gan, and Song Han. Efficientvit: Lightweight multi-scale attention for high-resolution dense prediction. InICCV, pages 17302– 17313, 2023. 3, 5

    2023

  4. [10]

    Retinexformer: One-stage retinex-based transformer for low-light image enhance- ment

    Yuanhao Cai, Hao Bian, Jing Lin, Haoqian Wang, Radu Timofte, and Yulun Zhang. Retinexformer: One-stage retinex-based transformer for low-light image enhance- ment. InICCV, pages 12504–12513, 2023. 7

    2023

  5. [11]

    Seeing motion in the dark

    Chen Chen, Qifeng Chen, Minh N Do, and Vladlen Koltun. Seeing motion in the dark. InICCV, pages 3185–3194,

  6. [12]

    Pre-trained image processing transformer

    Hanting Chen, Yunhe Wang, Tianyu Guo, Chang Xu, Yip- ing Deng, Zhenhua Liu, Siwei Ma, Chunjing Xu, Chao Xu, and Wen Gao. Pre-trained image processing transformer. In CVPR, pages 12299–12310, 2021. 2, 5, 7

    2021

  7. [13]

    Simple baselines for image restoration

    Liangyu Chen, Xiaojie Chu, Xiangyu Zhang, and Jian Sun. Simple baselines for image restoration. InECCV, pages 17–33, 2022. 4, 7

    2022

  8. [14]

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

    Wei-Ting Chen, Hao-Yu Fang, Jian-Jiun Ding, Cheng-Che Tsai, and Sy-Yen Kuo. Jstasr: Joint size and transparency- aware snow removal algorithm based on modified partial convolution and veiling effect removal. InECCV, pages 754–770, 2020. 7

    2020

  9. [15]

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

    Wei-Ting Chen, Hao-Yu Fang, Cheng-Lin Hsieh, Cheng- Che Tsai, I Chen, Jian-Jiun Ding, Sy-Yen Kuo, et al. All snow removed: Single image desnowing algorithm using hierarchical dual-tree complex wavelet representation and contradict channel loss. InICCV, pages 4196–4205, 2021. 7

    2021

  10. [16]

    Learning a sparse transformer network for effective image deraining

    Xiang Chen, Hao Li, Mingqiang Li, and Jinshan Pan. Learning a sparse transformer network for effective image deraining. InCVPR, pages 5896–5905, 2023. 1

    2023

  11. [17]

    Activating more pixels in image super- resolution transformer

    Xiangyu Chen, Xintao Wang, Jiantao Zhou, Yu Qiao, and Chao Dong. Activating more pixels in image super- resolution transformer. InCVPR, pages 22367–22377,

  12. [18]

    Dual aggregation transformer for image super-resolution

    Zheng Chen, Yulun Zhang, Jinjin Gu, Linghe Kong, Xi- aokang Yang, and Fisher Yu. Dual aggregation transformer for image super-resolution. InICCV, pages 12312–12321,

  13. [19]

    Hierarchical integration diffusion model for realistic image deblurring

    Zheng Chen, Yulun Zhang, Ding Liu, Jinjin Gu, Linghe Kong, Xin Yuan, et al. Hierarchical integration diffusion model for realistic image deblurring. InNeurIPS, 2023. 5, 6

    2023

  14. [20]

    Rethinking coarse-to-fine approach in single image deblurring

    Sung-Jin Cho, Seo-Won Ji, Jun-Pyo Hong, Seung-Won Jung, and Sung-Jea Ko. Rethinking coarse-to-fine approach in single image deblurring. InICCV, pages 4641–4650,

  15. [21]

    Fast and accurate deep network learn- ing by exponential linear units (elus).arXiv preprint arXiv:1511.07289, 2015

    Djork-Arn ´e Clevert. Fast and accurate deep network learn- ing by exponential linear units (elus).arXiv preprint arXiv:1511.07289, 2015. 2

    2015 arXiv

  16. [22]

    Focal network for image restoration

    Yuning Cui, Wenqi Ren, Xiaochun Cao, and Alois Knoll. Focal network for image restoration. InICCV, pages 13001–13011, 2023. 6, 7

    2023

  17. [23]

    Irnext: Rethinking convolutional network de- sign for image restoration

    Yuning Cui, Wenqi Ren, Sining Yang, Xiaochun Cao, and Alois Knoll. Irnext: Rethinking convolutional network de- sign for image restoration. InICML, 2023. 5

    2023

  18. [24]

    Language modeling with gated convolutional net- works

    Yann N Dauphin, Angela Fan, Michael Auli, and David Grangier. Language modeling with gated convolutional net- works. InICLR, pages 933–941, 2017. 5

    2017

  19. [25]

    Image super-resolution using deep convolutional net- works.TPAMI, 38(2):295–307, 2015

    Chao Dong, Chen Change Loy, Kaiming He, and Xiaoou Tang. Image super-resolution using deep convolutional net- works.TPAMI, 38(2):295–307, 2015. 1, 2

    2015

  20. [26]

    Multi-scale boosted de- hazing network with dense feature fusion

    Hang Dong, Jinshan Pan, Lei Xiang, Zhe Hu, Xinyi Zhang, Fei Wang, and Ming-Hsuan Yang. Multi-scale boosted de- hazing network with dense feature fusion. InCVPR, pages 2157–2167, 2020. 6

    2020

  21. [27]

    An image is worth 16x16 words: Transformers for image recognition at scale.arXiv preprint arXiv:2010.11929, 2020

    Alexey Dosovitskiy. An image is worth 16x16 words: Transformers for image recognition at scale.arXiv preprint arXiv:2010.11929, 2020. 2

    2010 arXiv

  22. [28]

    Cycle- dehaze: Enhanced cyclegan for single image dehazing

    Deniz Engin, Anil Genc ¸, and Hazim Kemal Ekenel. Cycle- dehaze: Enhanced cyclegan for single image dehazing. In CVPRW, pages 825–833, 2018. 7 9

    2018

  23. [29]

    Dy- namic scene deblurring with parameter selective sharing and nested skip connections

    Hongyun Gao, Xin Tao, Xiaoyong Shen, and Jiaya Jia. Dy- namic scene deblurring with parameter selective sharing and nested skip connections. InCVPR, pages 3848–3856,

  24. [30]

    Super-resolution in medical imaging.The computer journal, 52(1):43–63, 2009

    Hayit Greenspan. Super-resolution in medical imaging.The computer journal, 52(1):43–63, 2009. 1

    2009

  25. [31]

    Image dehazing transformer with transmission-aware 3d position embedding

    Chun-Le Guo, Qixin Yan, Saeed Anwar, Runmin Cong, Wenqi Ren, and Chongyi Li. Image dehazing transformer with transmission-aware 3d position embedding. InCVPR, pages 5812–5820, 2022. 6, 7

    2022

  26. [32]

    Flatten transformer: Vision transformer using focused linear attention

    Dongchen Han, Xuran Pan, Yizeng Han, Shiji Song, and Gao Huang. Flatten transformer: Vision transformer using focused linear attention. InICCV, pages 5961–5971, 2023. 3

    2023

  27. [33]

    Single image haze removal using dark channel prior.TPAMI, 33(12):2341– 2353, 2010

    Kaiming He, Jian Sun, and Xiaoou Tang. Single image haze removal using dark channel prior.TPAMI, 33(12):2341– 2353, 2010. 6

    2010

  28. [34]

    Image-to-image translation with conditional adver- sarial networks

    Phillip Isola, Jun-Yan Zhu, Tinghui Zhou, and Alexei A Efros. Image-to-image translation with conditional adver- sarial networks. InCVPR, 2017. 7

    2017

  29. [35]

    Edge-based defocus blur estimation with adaptive scale selection.TIP, 2017

    Ali Karaali and Claudio Rosito Jung. Edge-based defocus blur estimation with adaptive scale selection.TIP, 2017. 6

    2017

  30. [36]

    Transformers are rnns: Fast autore- gressive transformers with linear attention

    Angelos Katharopoulos, Apoorv Vyas, Nikolaos Pappas, and Franc ¸ois Fleuret. Transformers are rnns: Fast autore- gressive transformers with linear attention. InICML, pages 5156–5165, 2020. 2, 3, 4, 8

    2020

  31. [37]

    Accu- rate image super-resolution using very deep convolutional networks

    Jiwon Kim, Jung Kwon Lee, and Kyoung Mu Lee. Accu- rate image super-resolution using very deep convolutional networks. InCVPR, pages 1646–1654, 2016. 2

    2016

  32. [38]

    Mssnet: Multi-scale-stage network for single image deblurring

    Kiyeon Kim, Seungyong Lee, and Sunghyun Cho. Mssnet: Multi-scale-stage network for single image deblurring. In ECCV, pages 524–539, 2022. 5

    2022

  33. [39]

    Efficient frequency domain-based trans- formers for high-quality image deblurring

    Lingshun Kong, Jiangxin Dong, Jianjun Ge, Mingqiang Li, and Jinshan Pan. Efficient frequency domain-based trans- formers for high-quality image deblurring. InCVPR, pages 5886–5895, 2023. 5

    2023

  34. [40]

    Deblurgan: Blind mo- tion deblurring using conditional adversarial networks

    Orest Kupyn, V olodymyr Budzan, Mykola Mykhailych, Dmytro Mishkin, and Ji ˇr´ı Matas. Deblurgan: Blind mo- tion deblurring using conditional adversarial networks. In CVPR, pages 8183–8192, 2018. 5

    2018

  35. [41]

    Deblurgan-v2: Deblurring (orders-of- magnitude) faster and better

    Orest Kupyn, Tetiana Martyniuk, Junru Wu, and Zhangyang Wang. Deblurgan-v2: Deblurring (orders-of- magnitude) faster and better. InICCV, pages 8878–8887,

  36. [42]

    Photo-realistic single image super-resolution using a gen- erative adversarial network

    Christian Ledig, Lucas Theis, Ferenc Husz ´ar, Jose Ca- ballero, Andrew Cunningham, Alejandro Acosta, Andrew Aitken, Alykhan Tejani, Johannes Totz, Zehan Wang, et al. Photo-realistic single image super-resolution using a gen- erative adversarial network. InCVPR, pages 4681–4690,

  37. [43]

    Deep defocus map estimation using domain adapta- tion

    Junyong Lee, Sungkil Lee, Sunghyun Cho, and Seungyong Lee. Deep defocus map estimation using domain adapta- tion. InCVPR, 2019. 6

    2019

  38. [44]

    Iterative filter adaptive network for single image defocus deblurring

    Junyong Lee, Hyeongseok Son, Jaesung Rim, Sunghyun Cho, and Seungyong Lee. Iterative filter adaptive network for single image defocus deblurring. InCVPR, pages 2034– 2042, 2021. 6

    2021

  39. [45]

    Aod-net: All-in-one dehazing network

    Boyi Li, Xiulian Peng, Zhangyang Wang, Jizheng Xu, and Dan Feng. Aod-net: All-in-one dehazing network. InICCV, pages 4770–4778, 2017. 6

    2017

  40. [46]

    Benchmarking single-image dehazing and beyond.TIP, 28(1):492–505,

    Boyi Li, Wenqi Ren, Dengpan Fu, Dacheng Tao, Dan Feng, Wenjun Zeng, and Zhangyang Wang. Benchmarking single-image dehazing and beyond.TIP, 28(1):492–505,

  41. [47]

    All-in-one image restoration for unknown corruption

    Boyun Li, Xiao Liu, Peng Hu, Zhongqin Wu, Jiancheng Lv, and Xi Peng. All-in-one image restoration for unknown corruption. InCVPR, pages 17452–17462, 2022. 8

    2022

  42. [48]

    All in one bad weather removal using architectural search

    Ruoteng Li, Robby T Tan, and Loong-Fah Cheong. All in one bad weather removal using architectural search. In CVPR, pages 3175–3185, 2020. 7

    2020

  43. [49]

    Effi- cient and explicit modelling of image hierarchies for image restoration

    Yawei Li, Yuchen Fan, Xiaoyu Xiang, Denis Demandolx, Rakesh Ranjan, Radu Timofte, and Luc Van Gool. Effi- cient and explicit modelling of image hierarchies for image restoration. InCVPR, pages 18278–18289, 2023. 1, 2, 4, 5

    2023

  44. [50]

    Swinir: Image restoration us- ing swin transformer

    Jingyun Liang, Jiezhang Cao, Guolei Sun, Kai Zhang, Luc Van Gool, and Radu Timofte. Swinir: Image restoration us- ing swin transformer. InICCVW, pages 1833–1844, 2021. 1, 2, 7, 8

    2021

  45. [51]

    Enhanced deep residual networks for sin- gle image super-resolution

    Bee Lim, Sanghyun Son, Heewon Kim, Seungjun Nah, and Kyoung Mu Lee. Enhanced deep residual networks for sin- gle image super-resolution. InCVPRW, pages 136–144,

  46. [52]

    Residual denoising diffu- sion models

    Jiawei Liu, Qiang Wang, Huijie Fan, Yinong Wang, Yan- dong Tang, and Liangqiong Qu. Residual denoising diffu- sion models. InCVPR, pages 2773–2783, 2024. 8

    2024

  47. [53]

    Retinex-inspired unrolling with cooperative prior architecture search for low-light image enhancement

    Risheng Liu, Long Ma, Jiaao Zhang, Xin Fan, and Zhongx- uan Luo. Retinex-inspired unrolling with cooperative prior architecture search for low-light image enhancement. In CVPR, pages 10561–10570, 2021. 7

    2021

  48. [54]

    Griddehazenet: Attention-based multi-scale network for image dehazing

    Xiaohong Liu, Yongrui Ma, Zhihao Shi, and Jun Chen. Griddehazenet: Attention-based multi-scale network for image dehazing. InICCV, pages 7314–7323, 2019. 6

    2019

  49. [55]

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

    Xing Liu, Masanori Suganuma, Zhun Sun, and Takayuki Okatani. Dual residual networks leveraging the potential of paired operations for image restoration. InCVPR, pages 7007–7016, 2019. 7

    2019

  50. [56]

    Low-light image enhancement with multi-stage residue quantization and brightness-aware attention

    Yunlong Liu, Tao Huang, Weisheng Dong, Fangfang Wu, Xin Li, and Guangming Shi. Low-light image enhancement with multi-stage residue quantization and brightness-aware attention. InICCV, pages 12140–12149, 2023. 7

    2023

  51. [57]

    Desnownet: Context-aware deep network for snow removal.TIP, 27(6):3064–3073, 2018

    Yun-Fu Liu, Da-Wei Jaw, Shih-Chia Huang, and Jenq-Neng Hwang. Desnownet: Context-aware deep network for snow removal.TIP, 27(6):3064–3073, 2018. 7

    2018

  52. [58]

    Swin transformer: Hierarchical vision transformer using shifted windows

    Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin transformer: Hierarchical vision transformer using shifted windows. In ICCV, pages 10012–10022, 2021. 2

    2021

  53. [59]

    Controlling vision- language models for universal image restoration.arXiv preprint arXiv:2310.01018, 2023

    Ziwei Luo, Fredrik K Gustafsson, Zheng Zhao, Jens Sj¨olund, and Thomas B Sch ¨on. Controlling vision- language models for universal image restoration.arXiv preprint arXiv:2310.01018, 2023. 8

    2023 arXiv

  54. [60]

    Image restoration with mean-reverting stochastic differential equations

    Ziwei Luo, Fredrik K Gustafsson, Zheng Zhao, Jens Sj¨olund, and Thomas B Sch ¨on. Image restoration with mean-reverting stochastic differential equations. InICML, pages 23045–23066, 2023. 8 10

    2023

  55. [61]

    Prores: Explor- ing degradation-aware visual prompt for universal image restoration.arXiv preprint arXiv:2306.13653, 2023

    Jiaqi Ma, Tianheng Cheng, Guoli Wang, Qian Zhang, Xinggang Wang, and Lefei Zhang. Prores: Explor- ing degradation-aware visual prompt for universal image restoration.arXiv preprint arXiv:2306.13653, 2023. 8

    2023 arXiv

  56. [62]

    Toward fast, flexible, and robust low-light image enhancement

    Long Ma, Tengyu Ma, Risheng Liu, Xin Fan, and Zhongx- uan Luo. Toward fast, flexible, and robust low-light image enhancement. InCVPR, pages 5637–5646, 2022. 7

    2022

  57. [63]

    What can help pedestrian detection? InCVPR, pages 3127– 3136, 2017

    Jiayuan Mao, Tete Xiao, Yuning Jiang, and Zhimin Cao. What can help pedestrian detection? InCVPR, pages 3127– 3136, 2017. 2

    2017

  58. [64]

    Deep residual fourier transformation for single im- age deblurring.arXiv preprint arXiv:2111.11745, 2021

    Xintian Mao, Yiming Liu, Wei Shen, Qingli Li, and Yan Wang. Deep residual fourier transformation for single im- age deblurring.arXiv preprint arXiv:2111.11745, 2021. 5

    2021 arXiv

  59. [65]

    Image super-resolution with cross-scale non-local attention and exhaustive self- exemplars mining

    Yiqun Mei, Yuchen Fan, Yuqian Zhou, Lichao Huang, Thomas S Huang, and Honghui Shi. Image super-resolution with cross-scale non-local attention and exhaustive self- exemplars mining. InCVPR, pages 5690–5699, 2020. 2

    2020

  60. [66]

    Image super- resolution with non-local sparse attention

    Yiqun Mei, Yuchen Fan, and Yuqian Zhou. Image super- resolution with non-local sparse attention. InCVPR, pages 3517–3526, 2021. 2

    2021

  61. [67]

    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. InCVPR, pages 3883–3891, 2017. 5, 6, 7, 8

    2017

  62. [68]

    Restoring vision in adverse weather conditions with patch-based denoising diffusion models.TPAMI, 45(8):10346–10357, 2023

    Ozan ¨Ozdenizci and Robert Legenstein. Restoring vision in adverse weather conditions with patch-based denoising diffusion models.TPAMI, 45(8):10346–10357, 2023. 7, 8

    2023

  63. [69]

    Multi-temporal recurrent neural networks for pro- gressive non-uniform single image deblurring with incre- mental temporal training

    Dongwon Park, Dong Un Kang, Jisoo Kim, and Se Young Chun. Multi-temporal recurrent neural networks for pro- gressive non-uniform single image deblurring with incre- mental temporal training. InECCV, pages 327–343, 2020. 5

    2020

  64. [70]

    Scalable diffusion mod- els with transformers

    William Peebles and Saining Xie. Scalable diffusion mod- els with transformers. InICCV, pages 4195–4205, 2023. 3

    2023

  65. [71]

    Promptir: Prompting for all-in-one blind image restoration.arXiv preprint arXiv:2306.13090,

    Vaishnav Potlapalli, Syed Waqas Zamir, Salman Khan, and Fahad Shahbaz Khan. Promptir: Prompting for all-in-one blind image restoration.arXiv preprint arXiv:2306.13090,

    arXiv

  66. [72]

    Attentive generative adversarial network for rain- drop removal from a single image

    Rui Qian, Robby T Tan, Wenhan Yang, Jiajun Su, and Jiay- ing Liu. Attentive generative adversarial network for rain- drop removal from a single image. InCVPR, pages 2482– 2491, 2018. 7

    2018

  67. [73]

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

    Xu Qin, Zhilin Wang, Yuanchao Bai, Xiaodong Xie, and Huizhu Jia. Ffa-net: Feature fusion attention network for single image dehazing. InAAAI, pages 11908–11915, 2020. 6

    2020

  68. [74]

    Mb-taylorformer: Multi-branch efficient transformer expanded by taylor formula for image dehazing

    Yuwei Qiu, Kaihao Zhang, Chenxi Wang, Wenhan Luo, Hongdong Li, and Zhi Jin. Mb-taylorformer: Multi-branch efficient transformer expanded by taylor formula for image dehazing. InICCV, pages 12802–12813, 2023. 6, 7

    2023

  69. [75]

    Deep learning for seeing through window with raindrops

    Yuhui Quan, Shijie Deng, Yixin Chen, and Hui Ji. Deep learning for seeing through window with raindrops. In ICCV, pages 2463–2471, 2019. 7

    2019

  70. [76]

    Neumann network with recursive kernels for single image defocus deblurring

    Yuhui Quan, Zicong Wu, and Hui Ji. Neumann network with recursive kernels for single image defocus deblurring. InCVPR, pages 5754–5763, 2023. 6

    2023

  71. [77]

    Single image defocus de- blurring via implicit neural inverse kernels

    Yuhui Quan, Xin Yao, and Hui Ji. Single image defocus de- blurring via implicit neural inverse kernels. InICCV, pages 12600–12610, 2023. 6

    2023

  72. [78]

    Convolutional neural network super resolution for face recognition in surveillance monitoring

    Pejman Rasti, Tonis Uiboupin, Sergio Escalera, and Gho- lamreza Anbarjafari. Convolutional neural network super resolution for face recognition in surveillance monitoring. InArticulated Motion and Deformable Objects, pages 175– 184, 2016. 1

    2016

  73. [79]

    Real-world blur dataset for learning and benchmark- ing deblurring algorithms

    Jaesung Rim, Haeyun Lee, Jucheol Won, and Sunghyun Cho. Real-world blur dataset for learning and benchmark- ing deblurring algorithms. InECCV, pages 184–201, 2020. 5

    2020

  74. [80]

    Learning to deblur using light field generated and real de- focus images

    Lingyan Ruan, Bin Chen, Jizhou Li, and Miuling Lam. Learning to deblur using light field generated and real de- focus images. InCVPR, pages 16304–16313, 2022. 6

    2022

  75. [81]

    Human-aware mo- tion deblurring

    Ziyi Shen, Wenguan Wang, Xiankai Lu, Jianbing Shen, Haibin Ling, Tingfa Xu, and Ling Shao. Human-aware mo- tion deblurring. InICCV, pages 5572–5581, 2019. 5

    2019

  76. [82]

    Efficient attention: Attention with lin- ear complexities

    Zhuoran Shen, Mingyuan Zhang, Haiyu Zhao, Shuai Yi, and Hongsheng Li. Efficient attention: Attention with lin- ear complexities. InWACV, pages 3531–3539, 2021. 3

    2021

  77. [83]

    Just noticeable defocus blur detection and estimation

    Jianping Shi, Li Xu, and Jiaya Jia. Just noticeable defocus blur detection and estimation. InCVPR, 2015. 6

    2015

  78. [84]

    Single image defocus deblurring using kernel-sharing parallel atrous convolutions

    Hyeongseok Son, Junyong Lee, Sunghyun Cho, and Se- ungyong Lee. Single image defocus deblurring using kernel-sharing parallel atrous convolutions. InICCV, pages 2642–2650, 2021. 6

    2021

  79. [85]

    Spatially-attentive patch-hierarchical network for adaptive motion deblurring

    Maitreya Suin, Kuldeep Purohit, and AN Rajagopalan. Spatially-attentive patch-hierarchical network for adaptive motion deblurring. InCVPR, pages 3606–3615, 2020. 5

    2020

  80. [86]

    Restoring images in adverse weather con- ditions via histogram transformer

    Shangquan Sun, Wenqi Ren, Xinwei Gao, Rui Wang, and Xiaochun Cao. Restoring images in adverse weather con- ditions via histogram transformer. InECCV, 2024. 1, 7

    2024

  81. [87]

    Scale-recurrent network for deep image deblurring

    Xin Tao, Hongyun Gao, Xiaoyong Shen, Jue Wang, and Ji- aya Jia. Scale-recurrent network for deep image deblurring. InCVPR, pages 8174–8182, 2018. 5

    2018

  82. [88]

    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. InECCV, pages 146–162, 2022. 5

    2022

  83. [89]

    Banet: a blur-aware attention net- work for dynamic scene deblurring.TIP, 31:6789–6799,

    Fu-Jen Tsai, Yan-Tsung Peng, Chung-Chi Tsai, Yen-Yu Lin, and Chia-Wen Lin. Banet: a blur-aware attention net- work for dynamic scene deblurring.TIP, 31:6789–6799,

  84. [90]

    Maxim: Multi-axis mlp for image processing

    Zhengzhong Tu, Hossein Talebi, Han Zhang, Feng Yang, Peyman Milanfar, Alan Bovik, and Yinxiao Li. Maxim: Multi-axis mlp for image processing. InCVPR, pages 5769–5780, 2022. 6, 7

    2022

  85. [91]

    Transweather: Transformer-based restoration of im- ages degraded by adverse weather conditions

    Jeya Maria Jose Valanarasu, Rajeev Yasarla, and Vishal M Patel. Transweather: Transformer-based restoration of im- ages degraded by adverse weather conditions. InCVPR, pages 2353–2363, 2022. 7

    2022

  86. [92]

    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. InNeurIPS, pages 5998–6008, 2017. 2, 3 11

    2017

  87. [94]

    Pyramid vision transformer: A versatile backbone for dense prediction without convolutions

    Wenhai Wang, Enze Xie, Xiang Li, Deng-Ping Fan, Kaitao Song, Ding Liang, Tong Lu, Ping Luo, and Ling Shao. Pyramid vision transformer: A versatile backbone for dense prediction without convolutions. InICCV, pages 568–578,

  88. [95]

    Esrgan: Enhanced super-resolution generative adversarial networks.arXiv preprint arXiv:1809.00219, 2018

    Xintao Wang, Ke Yu, Shixiang Wu, Jinjin Gu, Yihao Liu, Chao Dong, Chen Change Loy, Yu Qiao, and Xiaoou Tang. Esrgan: Enhanced super-resolution generative adversarial networks.arXiv preprint arXiv:1809.00219, 2018. 2

    2018 arXiv

  89. [96]

    Images speak in images: A generalist painter for in-context visual learning

    Xinlong Wang, Wen Wang, Yue Cao, Chunhua Shen, and Tiejun Huang. Images speak in images: A generalist painter for in-context visual learning. InCVPR, pages 6830–6839,

  90. [97]

    Low-light image enhancement with normalizing flow

    Yufei Wang, Renjie Wan, Wenhan Yang, Haoliang Li, Lap- Pui Chau, and Alex Kot. Low-light image enhancement with normalizing flow. InAAAI, pages 2604–2612, 2022. 7

    2022

  91. [98]

    Uformer: A gen- eral u-shaped transformer for image restoration

    Zhendong Wang, Xiaodong Cun, Jianmin Bao, Wengang Zhou, Jianzhuang Liu, and Houqiang Li. Uformer: A gen- eral u-shaped transformer for image restoration. InCVPR, pages 17683–17693, 2022. 1, 2, 4, 5, 6, 7

    2022

  92. [99]

    Deep retinex decomposition for low-light enhancement

    Chen Wei, Wenjing Wang, Wenhan Yang, and Jiaying Liu. Deep retinex decomposition for low-light enhancement. arXiv preprint arXiv:1808.04560, 2018. 7

    2018 arXiv

  93. [100]

    Con- trastive learning for compact single image dehazing

    Haiyan Wu, Yanyun Qu, Shaohui Lin, Jian Zhou, Ruizhi Qiao, Zhizhong Zhang, Yuan Xie, and Lizhuang Ma. Con- trastive learning for compact single image dehazing. In CVPR, pages 10551–10560, 2021. 6

    2021

  94. [101]

    Uretinex-net: Retinex-based deep unfolding network for low-light image enhancement

    Wenhui Wu, Jian Weng, Pingping Zhang, Xu Wang, Wen- han Yang, and Jianmin Jiang. Uretinex-net: Retinex-based deep unfolding network for low-light image enhancement. InCVPR, pages 5901–5910, 2022. 7

    2022

  95. [102]

    Efficient non-local contrastive attention for image super-resolution

    Bin Xia, Yucheng Hang, Yapeng Tian, Wenming Yang, Qingmin Liao, and Jie Zhou. Efficient non-local contrastive attention for image super-resolution. InAAAI, pages 2759– 2767, 2022. 2

    2022

  96. [103]

    Diffir: Efficient diffusion model for image restoration

    Bin Xia, Yulun Zhang, Shiyin Wang, Yitong Wang, Xing- long Wu, Yapeng Tian, Wenming Yang, and Luc Van Gool. Diffir: Efficient diffusion model for image restoration. In ICCV, pages 13095–13105, 2023. 2

    2023

  97. [104]

    Image de-raining transformer.TPAMI, 45(11): 12978–12995, 2023

    Jie Xiao, Xueyang Fu, Aiping Liu, Feng Wu, and Zheng- Jun Zha. Image de-raining transformer.TPAMI, 45(11): 12978–12995, 2023. 1, 7

    2023

  98. [105]

    Sana: Efficient high-resolution image synthesis with linear diffu- sion transformers.arXiv preprint arXiv:2410.10629, 2024

    Enze Xie, Junsong Chen, Junyu Chen, Han Cai, Yujun Lin, Zhekai Zhang, Muyang Li, Yao Lu, and Song Han. Sana: Efficient high-resolution image synthesis with linear diffu- sion transformers.arXiv preprint arXiv:2410.10629, 2024. 3

    arXiv

  99. [106]

    Deep convolu- tional neural network for image deconvolution

    Li Xu, Jimmy S Ren, Ce Liu, and Jiaya Jia. Deep convolu- tional neural network for image deconvolution. InNeurIPS,

  100. [107]

    Snr-aware low-light image enhancement

    Xiaogang Xu, Ruixing Wang, Chi-Wing Fu, and Jiaya Jia. Snr-aware low-light image enhancement. InCVPR, pages 17714–17724, 2022. 7

    2022

  101. [108]

    From fidelity to perceptual quality: A semi- supervised approach for low-light image enhancement

    Wenhan Yang, Shiqi Wang, Yuming Fang, Yue Wang, and Jiaying Liu. From fidelity to perceptual quality: A semi- supervised approach for low-light image enhancement. In CVPR, pages 3063–3072, 2020. 7

    2020

  102. [109]

    Sparse gradient regularized deep retinex network for robust low-light image enhancement

    Wenhan Yang, Wenjing Wang, Haofeng Huang, Shiqi Wang, and Jiaying Liu. Sparse gradient regularized deep retinex network for robust low-light image enhancement. TIP, 30:2072–2086, 2021. 7

    2021

  103. [110]

    Perceiving and modeling density for image dehazing

    Tian Ye, Yunchen Zhang, Mingchao Jiang, Liang Chen, Yun Liu, Sixiang Chen, and Erkang Chen. Perceiving and modeling density for image dehazing. InECCV, pages 130– 145, 2022. 6

    2022

  104. [111]

    Ad- verse weather removal with codebook priors

    Tian Ye, Sixiang Chen, Jinbin Bai, Jun Shi, Chenghao Xue, Jingxia Jiang, Junjie Yin, Erkang Chen, and Yun Liu. Ad- verse weather removal with codebook priors. InICCV,

  105. [112]

    Learning diffusion texture priors for image restoration

    Tian Ye, Sixiang Chen, Wenhao Chai, Zhaohu Xing, Jing Qin, Ge Lin, and Lei Zhu. Learning diffusion texture priors for image restoration. InCVPR, pages 2524–2534, 2024. 7

    2024

  106. [113]

    Learning enriched features for real image restoration and enhancement

    Syed Waqas Zamir, Aditya Arora, Salman Khan, Munawar Hayat, Fahad Shahbaz Khan, Ming-Hsuan Yang, and Ling Shao. Learning enriched features for real image restoration and enhancement. InECCV, pages 492–511, 2020. 7

    2020

  107. [114]

    Multi-stage progressive image restoration

    Syed Waqas Zamir, Aditya Arora, Salman Khan, Munawar Hayat, Fahad Shahbaz Khan, Ming-Hsuan Yang, and Ling Shao. Multi-stage progressive image restoration. InCVPR, pages 14821–14831, 2021. 5, 6

    2021

  108. [116]

    Learning enriched features for fast image restoration and enhancement.TPAMI, 45(2):1934–1948, 2022

    Syed Waqas Zamir, Aditya Arora, Salman Khan, Munawar Hayat, Fahad Shahbaz Khan, Ming-Hsuan Yang, and Ling Shao. Learning enriched features for fast image restoration and enhancement.TPAMI, 45(2):1934–1948, 2022. 8

    1934

  109. [117]

    Deep stacked hierarchical multi-patch network for image deblurring

    Hongguang Zhang, Yuchao Dai, Hongdong Li, and Piotr Koniusz. Deep stacked hierarchical multi-patch network for image deblurring. InCVPR, pages 5978–5986, 2019. 5

    2019

  110. [118]

    Dy- namic scene deblurring using spatially variant recurrent neural networks

    Jiawei Zhang, Jinshan Pan, Jimmy Ren, Yibing Song, Lin- chao Bao, Rynson WH Lau, and Ming-Hsuan Yang. Dy- namic scene deblurring using spatially variant recurrent neural networks. InCVPR, pages 2521–2529, 2018. 5

    2018

  111. [119]

    Ingredient-oriented multi- degradation learning for image restoration

    Jinghao Zhang, Jie Huang, Mingde Yao, Zizheng Yang, Hu Yu, Man Zhou, and Feng Zhao. Ingredient-oriented multi- degradation learning for image restoration. InCVPR, pages 5825–5835, 2023. 8

    2023

  112. [120]

    Accurate image restoration with attention retractable transformer

    Jiale Zhang, Yulun Zhang, Jinjin Gu, Yongbing Zhang, Linghe Kong, and Xin Yuan. Accurate image restoration with attention retractable transformer. InICLR, 2023. 1, 3

    2023

  113. [121]

    Beyond a gaussian denoiser: Residual learning of deep cnn for image denoising.TIP, 26(7):3142–3155,

    Kai Zhang, Wangmeng Zuo, Yunjin Chen, Deyu Meng, and Lei Zhang. Beyond a gaussian denoiser: Residual learning of deep cnn for image denoising.TIP, 26(7):3142–3155,

  114. [122]

    Deblurring by realistic blurring

    Kaihao Zhang, Wenhan Luo, Yiran Zhong, Lin Ma, Bjorn Stenger, Wei Liu, and Hongdong Li. Deblurring by realistic blurring. InCVPR, pages 2737–2746, 2020. 5 12

    2020

  115. [123]

    Hit-sr: Hierar- chical transformer for efficient image super-resolution

    Xiang Zhang, Yulun Zhang, and Fisher Yu. Hit-sr: Hierar- chical transformer for efficient image super-resolution. In ECCV, 2024. 2

    2024

  116. [124]

    Image super-resolution using very deep residual channel attention networks

    Yulun Zhang, Kunpeng Li, Kai Li, Lichen Wang, Bineng Zhong, and Yun Fu. Image super-resolution using very deep residual channel attention networks. InECCV, pages 286– 301, 2018. 2, 4

    2018

  117. [125]

    Residual dense network for image super- resolution

    Yulun Zhang, Yapeng Tian, Yu Kong, Bineng Zhong, and Yun Fu. Residual dense network for image super- resolution. InCVPR, pages 2472–2481, 2018. 2

    2018

  118. [126]

    Kin- dling the darkness: A practical low-light image enhancer

    Yonghua Zhang, Jiawan Zhang, and Xiaojie Guo. Kin- dling the darkness: A practical low-light image enhancer. InACMMM, pages 1632–1640, 2019. 7

    2019

  119. [127]

    Residual dense network for image restoration

    Yulun Zhang, Yapeng Tian, Yu Kong, Bineng Zhong, and Yun Fu. Residual dense network for image restoration. TPAMI, 43(7):2480–2495, 2021. 2

    2021

  120. [128]

    Selective hourglass mapping for universal image restoration based on diffusion model

    Dian Zheng, Xiao-Ming Wu, Shuzhou Yang, Jian Zhang, Jian-Fang Hu, and Wei-Shi Zheng. Selective hourglass mapping for universal image restoration based on diffusion model. InCVPR, pages 25445–25455, 2024. 7, 8

    2024

  121. [129]

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

    Shihao Zhou, Duosheng Chen, Jinshan Pan, Jinglei Shi, and Jufeng Yang. Adapt or perish: Adaptive sparse transformer with attentive feature refinement for image restoration. In CVPR, pages 2952–2963, 2024. 1, 3, 6, 7

    2024

  122. [130]

    Seeing the unseen: A fre- quency prompt guided transformer for image restoration

    Shihao Zhou, Jinshan Pan, Jinglei Shi, Duosheng Chen, Lishen Qu, and Jufeng Yang. Seeing the unseen: A fre- quency prompt guided transformer for image restoration. InECCV, pages 246–264, 2024. 5, 6

    2024

  123. [131]

    Traffic-sign detection and classification in the wild

    Zhe Zhu, Dun Liang, Songhai Zhang, Xiaolei Huang, Baoli Li, and Shimin Hu. Traffic-sign detection and classification in the wild. InCVPR, pages 2110–2118, 2016. 1, 2 13

    2016

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.