Underwater Image Restoration via Polymorphic Large Kernel CNNs

Chi-Man Pun; Fuchen Zheng; Weiwen Chen; Xiaojiao Guo; Xuhang Chen; Yihang Dong; Zimeng Li

REVIEW 4 major objections 5 minor 2 cited by

Underwater Image Restoration via Polymorphic Large Kernel CNNs

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A carefully designed pure CNN can match transformer-based models on underwater image restoration, the paper claims.

desk verdict A solid, lightweight CNN architecture for underwater restoration with a real design contribution, but the SOTA claim is weakened by the evaluation protocol: algorithm-generated references and UCIQE in the training loss. read the letter →

arxiv 2412.18459 v1 pith:5YXG6EGS submitted 2024-12-24 cs.CV eess.IV

Xiaojiao Guo , Yihang Dong , Xuhang Chen , Weiwen Chen , Zimeng Li , FuChen Zheng , Chi-Man Pun This is my paper · ORCID

classification cs.CVeess.IV

keywords underwaterimagerestorationlargekernelCNNcompositeshapeconvolutionhybriddomainattentionfrequencylightweightarchitectureenhancement

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper claims that a carefully built pure CNN, UIR-PolyKernel, matches or beats heavier transformer-based models on underwater image restoration benchmarks while using a fraction of the compute. The two load-bearing design choices are polymorphic large-kernel convolutions, which capture long-range and direction-dependent degradation cheaply, and hybrid frequency-space attention, which preserves fine detail. On paired test sets from UIEB, EUVP, and LSUI and unpaired RUIE data, the model reports leading SSIM, LPIPS, and UCIQE scores, with a few exceptions such as second place to Semi-UIR on UIEB PSNR. A reader should care because it suggests that pure CNN architectures are not a compromise for this task, and that restoration quality can be obtained at 1.84M parameters and 13.67G MACs.

What carries the argument

The central mechanism is the Composite Shape Convolution (CSC) module: a parallel composition of depthwise strip, square, and pointwise convolutions with kernels 31x1, 1x31, 31x31, and 1x1, followed by a 1x1 pointwise convolution for inter-channel mixing. It is placed at the 64x64 bottleneck of a U-Net, where a single 31x31 depthwise convolution over the whole feature map would be costly; the polymorphic shapes reduce that cost while preserving both isotropic and anisotropic long-range context. Around it, Large Kernel Attention (LKA) modules simulate roughly 19x19 receptive fields for medium-range relations, and the Hybrid Domain Attention (HDA) module uses a Fast Fourier Transform branch so that attention can act on frequency components that have no obvious spatial-domain counterpart.

What would settle it

Take a set of underwater scenes photographed with known color-calibration targets at measured depths, use the measured target values as true references, and recompute PSNR, SSIM, and LPIPS for UIR-PolyKernel and the transformer baselines; if the margin over the baselines disappears or reverses, the reported superiority is an artifact of algorithm-generated references.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is that a purely convolutional encoder-bottleneck-decoder, UIR-PolyKernel, can reach state-of-the-art underwater restoration results. At the bottleneck, the Composite Shape Convolution module runs four parallel depthwise convolutions (31x1, 1x31, 31x31, and 1x1) on quarter-resolution features to model global context and anisotropic scattering; at two intermediate scales, Large Kernel Attention modules extend receptive fields; at full resolution, the Hybrid Domain Attention module applies an FFT branch, pixel attention, and channel attention. Trained on 4800 real underwater images with a loss combining SmoothL1, SSIM, and UCIQE, the model achieves the best or near-best PSNR, SSIM, LPIPS, and UCIQE across four benchmarks, including a 26.42 PSNR on EUVP, while the ablation study attributes the largest single gain to the frequency-domain FDPA block.

Load-bearing premise

The reported quality scores assume the reference images used for training and evaluation are trustworthy stand-ins for true clean underwater scenes, even though those references are themselves outputs of other algorithms rather than measured ground truth.

Editorial extensions

If this is right

UIR-PolyKernel's per-image cost of 13.67 G MACs and 1.84M parameters makes near-state-of-the-art restoration feasible on edge devices.
The design suggests that large, heterogeneous kernel shapes can replace self-attention for long-range dependencies in low-level vision.
Frequency-domain pixel attention can expose degradation cues invisible in the spatial domain, improving robustness across different water types.
The composite loss that includes UCIQE nudges outputs toward perceptually good underwater quality beyond pixel fidelity.

Reading between the lines

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

Because the benchmark references are themselves algorithm-generated outputs, the stated PSNR, SSIM, and LPIPS gains may measure fidelity to those machine targets rather than to true clean scenes; a physically grounded reference benchmark could change the ranking.
The same polymorphic-kernel and frequency-spatial attention recipe should transfer to other degradation tasks such as dehazing, deraining, or low-light enhancement, where anisotropic degradation and subtle frequency cues also dominate.
A straightforward extension would be a human perceptual study on the restored images; if human preference tracks the reported metrics, the method's practical value is confirmed despite the imperfect references.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

The paper is worth a look if you work in underwater image restoration. The architecture is a genuine combination, not a paradigm shift: polymorphic depth-wise convolutions (31x1, 1x31, 31x31) at the bottleneck, LKA modules at medium scales, and a hybrid frequency/spatial attention block at full resolution. The ablations are clean and consistent, each module adds a measurable gain, and the baselines were retrained under the same data, which is more than many papers in this area do. At 1.84M params and 13.67 GMACs it is genuinely light, and it beats heavier Transformer-based models on most metrics on the paired test sets. That is a real result for practitioners who care about deployment cost. The soft spots are real but not fatal. The paper itself says in Section II-D that the reference targets are not true ground truth, only outputs of existing methods. So the PSNR/SSIM/LPIPS numbers measure how well the model reproduces a particular set of algorithm-generated references, not actual restoration quality. That alone should temper the state-of-the-art claim. Second, UCIQE appears both in the loss (Eq. 3, w3=0.01) and as a headline success metric; the reported UCIQE advantage is partly a fitted quantity. The weight is small, so this is a minor-to-moderate issue, but it should be acknowledged or the metric dropped from the loss. Third, there are no error bars or repeated runs. On UIEB PSNR the method is second (23.52 vs 23.64 for Semi-UIR), so the large margin language is overstated. The paper is honest about the reference limitation, which counts in its favor. The math is straightforward and the code is public. The main weakness is that the empirical evidence underdetermines the grand claim. It is a good engineering contribution, not a breakthrough. Who should read it: anyone building lightweight underwater enhancement models, or benchmarking CNNs against Transformers in low-level vision. It deserves a serious referee, desk rejection would be wrong. I would recommend conditional acceptance, with the authors asked to either remove UCIQE from the loss, report results on a held-out set with true references where possible, or at minimum re-frame the SOTA claim to best among lightweight CNNs and provide error bars.

Referee Report

4 major / 5 minor

Summary. The paper proposes UIR-PolyKernel, a lightweight pure-CNN architecture for underwater image restoration. The method combines polymorphic large-kernel convolutions (Composite Shape Convolution module with 31x1, 1x31, 31x31, and 1x1 depth-wise convolutions at the bottleneck), Large Kernel Attention modules at intermediate scales, and a Hybrid Domain Attention module that fuses frequency-domain pixel attention with spatial-domain channel attention. The model has about 1.84M parameters and 13.67 GMACs. Experiments on UIEB, EUVP, LSUI, and RUIE report PSNR, SSIM, LPIPS, and UCIQE, with all deep-learning baselines retrained on the same training data. Ablations show each component contributes. The central claim is state-of-the-art quantitative and qualitative performance relative to existing methods, including Transformers.

Significance. If the claims are reliable, the paper makes a useful practical point: a carefully designed lightweight CNN can compete with heavier Transformer-based underwater restoration models. The architecture is coherent, the ablation study is systematic, all baselines are retrained under a common protocol, and source code is provided. However, the evaluation protocol has serious limitations that directly affect the headline claim. The paper itself states that the reference targets are not true ground truth, and UCIQE appears both as a training loss term and as a reported success metric. These issues, combined with the lack of error bars and a specific counterexample in the UIEB PSNR column, undermine the 'state-of-the-art' assertion. The underlying architectural ideas are plausible and worth further validation, but the current evidence is not sufficient to support the central claim.

major comments (4)

[Section II-D (Eq. 3) and Table I] The loss function in Eq. (3) includes the term w3 * Lu, where Lu = 1 - UCIQE and w3 = 0.01. This makes UCIQE a training objective. Table I then reports UCIQE as one of the four success metrics. Because the method is trained to increase UCIQE, the reported UCIQE advantage over competing methods (e.g., 0.458 on UIEB vs. 0.432 for URSCT) is at least partly an artifact of direct optimization rather than an independent quality measure. The paper itself cautions in Section III-C that UCIQE is 'not absolutely reliable,' yet still presents it as a headline result. The authors should either remove UCIQE from the training loss or not report UCIQE as a comparative benchmark metric.
[Section II-D and Table I] Section II-D explicitly states that 'the reference target, while consisting of high-quality results from state-of-the-art methods, does not represent the true ground truth.' Under this protocol, PSNR, SSIM, and LPIPS in Table I measure similarity to algorithm-generated reference outputs, not to true clean scene radiance. A model could achieve high scores by memorizing the statistics of these references without genuinely restoring underwater scenes. This is load-bearing for the 'state-of-the-art' claim, especially because all deep-learning baselines are retrained on the same references. The authors should provide independent validation on data with true ground truth (e.g., synthetic underwater scenes with known clean images) or include a human perceptual study to support the qualitative superiority asserted in Section III-C.
[Table I, UIEB PSNR] On the UIEB test set, the method achieves PSNR 23.52 dB, which is second to Semi-UIR's 23.64 dB. Section III-C states that 'In all other results, our method leads other approaches by a large margin,' but this is not true for a primary metric on the primary dataset. Moreover, no error bars, confidence intervals, or repeated runs are reported for any metric, so the margins on other datasets (e.g., EUVP PSNR 26.42 vs. 25.74 for URSCT) are not established as statistically significant. The authors should qualify the state-of-the-art claim and provide variance estimates or significance tests.
[Table II, Ablation Studies] The ablation results show incremental improvements from adding each module (e.g., SSIM changes from 0.888 to 0.891 to 0.893 when HDA components are added). Without repeated runs or statistical testing, it is unclear whether these differences exceed training noise. The conclusion that each component contributes to 'superior performance' is therefore not robust. The authors should report means and standard deviations over multiple random seeds, or at least perform significance tests on the differences.

minor comments (5)

[Section II-C, Eq. (2)] The text says the FDPA block has 'two 1x1 convolution branches,' but Eq. (2) uses the same notation Conv1 twice. Please clarify whether the two branches have separate learnable weights or share them.
[Section III-C] The text states 'we limit our UCIQE comparison to the deep learning methods,' yet Table I lists UCIQE values for all methods, including traditional ones. Please reconcile the statement with the table.
[Abstract] The phrase 'underwater and on-air images' appears to contain a typo; likely 'in-air' or 'atmospheric' images are intended.
[Section III-A] The UCIQE metric is mentioned without a citation. Please add a reference to the original UCIQE paper.
[Author affiliations] There are several spelling errors in affiliations ('Univeristy' appears twice). Please proofread the author information.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: benchmark evaluation is external and held-out; UCIQE-in-loss is a protocol caveat, not a fitted prediction.

full rationale

The paper's central claim is an empirical benchmark result, not a derivation from first principles. The architecture is presented as a combination of existing large-kernel and attention building blocks (OKNet, LSKA, VAN), with ablations (Table II) showing each module contributes on held-out test sets. The quantitative SOTA claim rests on Table I, where all deep learning methods are retrained on the same training data and evaluated on disjoint test splits (90 UIEB / 200 EUVP / 200 LSUI, plus unpaired RUIE). The loss in Eq. (3) includes w3·(1−UCIQE), and UCIQE is also reported in Table I; this is a mild protocol concern, but because the reported numbers are measured on held-out test images rather than on the training subset used to fit the weights, it does not constitute a fitted-input-called-prediction circularity. The paper explicitly discloses that the reference targets are algorithm-generated and do not represent true ground truth; this is a benchmark-validity caveat, not a circularity in the claimed derivation. No load-bearing self-citation or imported uniqueness theorem appears; citations to the authors' prior works (e.g., [20]–[25], [31], [32]) are contextual and do not carry the argument. Therefore no specific circular step is identified.

Assumptions & free parameters 3 free parameters · 2 assumptions · 0 invented entities

The central claim rests on standard supervised training with a tuned loss; no physical entities or novel mathematical axioms are introduced. Main hand-set quantities are loss weights, kernel size 31, and initial channel count.

free parameters (3)

Loss weights w1, w2, w3 = 1, 0.2, 0.01
Optimized through experimental procedures (Section II-D); they control the balance of pixel, SSIM, and UCIQE terms in the training loss.
Large kernel size in CSC module = 31
Selected experimentally from 15, 31, and 63 (Section II-A) as a compromise between receptive field and computational cost.
Initial number of channels = 36
Chosen as the base width of the first feature level (Fig. 1 caption); affects capacity and parameter count.

assumptions (2)

domain assumption The reference images in UIEB, EUVP, and LSUI are valid supervised targets, despite not being true ground truth.
Stated in Section II-D that the targets are outputs of high-quality methods; the entire training and evaluation assume these references are good enough proxies for clean images.
domain assumption UCIQE is a meaningful non-reference underwater quality metric.
Used in the loss function (Eq. 3) and as an evaluation metric (Section III-A), although the paper notes it is 'not absolutely reliable'.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Underwater Image Restoration via Polymorphic Large Kernel CNNs." pith.science (2026). https://pith.science/paper/5YXG6EGS

@misc{pith2026241218459,
  author       = {Pith},
  title        = {Pith review of: Underwater Image Restoration via Polymorphic Large Kernel CNNs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5YXG6EGS}},
  note         = {Machine review of arXiv:2412.18459}
}

read the original abstract

Underwater Image Restoration (UIR) remains a challenging task in computer vision due to the complex degradation of images in underwater environments. While recent approaches have leveraged various deep learning techniques, including Transformers and complex, parameter-heavy models to achieve significant improvements in restoration effects, we demonstrate that pure CNN architectures with lightweight parameters can achieve comparable results. In this paper, we introduce UIR-PolyKernel, a novel method for underwater image restoration that leverages Polymorphic Large Kernel CNNs. Our approach uniquely combines large kernel convolutions of diverse sizes and shapes to effectively capture long-range dependencies within underwater imagery. Additionally, we introduce a Hybrid Domain Attention module that integrates frequency and spatial domain attention mechanisms to enhance feature importance. By leveraging the frequency domain, we can capture hidden features that may not be perceptible to humans but are crucial for identifying patterns in both underwater and on-air images. This approach enhances the generalization and robustness of our UIR model. Extensive experiments on benchmark datasets demonstrate that UIR-PolyKernel achieves state-of-the-art performance in underwater image restoration tasks, both quantitatively and qualitatively. Our results show that well-designed pure CNN architectures can effectively compete with more complex models, offering a balance between performance and computational efficiency. This work provides new insights into the potential of CNN-based approaches for challenging image restoration tasks in underwater environments. The code is available at \href{https://github.com/CXH-Research/UIR-PolyKernel}{https://github.com/CXH-Research/UIR-PolyKernel}.

Figures

Figures reproduced from arXiv: 2412.18459 by the authors.

**Figure 1.** The overall architecture of the proposed UIR-PolyKernel model. This model utilizes three resolution levels of feature maps, generated by two pairs [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗

**Figure 2.** Visual comparison of the proposed method with two leading traditional approaches and the three top-performing deep learning methods. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

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

DEEP-SEA: Deep-Learning Enhancement for Environmental Perception in Submerged Aquatics
cs.CV 2025-08 unverdicted novelty 4.0 of 10

DEEP-SEA, a dual-frequency self-attention network, reports state-of-the-art underwater image restoration on the EUVP and LSUI benchmarks.
Moir\'e Zero: An Efficient and High-Performance Neural Architecture for Moir\'e Removal
cs.CV 2025-07 conditional novelty 4.0 of 10

MZNet reports state-of-the-art moiré removal on high-resolution benchmarks with a U-Net combining multi-scale dual attention, multi-shape large-kernel convolutions, and fused skip connections.

Reference graph

Works this paper leans on

47 extracted references · 46 canonical work pages · cited by 2 Pith papers

[1]

A variational framework for underwater image dehazing and deblurring,

J. Xie, G. Hou, G. Wang, and Z. Pan, “A variational framework for underwater image dehazing and deblurring,” TCSVT, vol. 32, no. 6, pp. 3514–3526, 2021

work page 2021
[2]

Under- water image enhancement via minimal color loss and locally adaptive contrast enhancement,

W. Zhang, P. Zhuang, H.-H. Sun, G. Li, S. Kwong, and C. Li, “Under- water image enhancement via minimal color loss and locally adaptive contrast enhancement,” TIP, vol. 31, pp. 3997–4010, 2022

work page 2022
[3]

Rank-one prior: Real-time scene recovery,

J. Liu, R. W. Liu, J. Sun, and T. Zeng, “Rank-one prior: Real-time scene recovery,” TPAMI, 2022

work page 2022
[4]

Underwater camera: Improving visual perception via adaptive dark pixel prior and color correction,

J. Zhou, Q. Liu, Q. Jiang, W. Ren, K.-M. Lam, and W. Zhang, “Underwater camera: Improving visual perception via adaptive dark pixel prior and color correction,” IJCV, pp. 1–19, 2023

work page 2023
[5]

Underwater image enhancement via weighted wavelet visual perception fusion,

W. Zhang, L. Zhou, P. Zhuang, G. Li, X. Pan, W. Zhao, and C. Li, “Underwater image enhancement via weighted wavelet visual perception fusion,” TCSVT, 2023

work page 2023
[6]

Sienet: Siamese expansion network for image extrapolation,

X. Zhang, F. Chen, C. Wang, M. Tao, and G.-P. Jiang, “Sienet: Siamese expansion network for image extrapolation,” SPL, vol. 27, pp. 1590– 1594, 2020

work page 2020
[7]

Shadclips: When parameter-efficient fine-tuning with multimodal meets shadow removal,

X. Zhang, Z. Xu, H. Tang, C. Gu, S. Zhu, and X. Guan, “Shadclips: When parameter-efficient fine-tuning with multimodal meets shadow removal,” 2024

work page 2024
[8]

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,” TIP, vol. 30, pp. 4985–5000, 2021

work page 2021

Show all 47 references

[9]

Beyond single reference for training: underwater image enhancement via comparative learning,

K. Li, L. Wu, Q. Qi, W. Liu, X. Gao, L. Zhou, and D. Song, “Beyond single reference for training: underwater image enhancement via comparative learning,” TCSVT, 2022

2022
[10]

Toward sufficient spatial- frequency interaction for gradient-aware underwater image enhance- ment,

C. Zhao, W. Cai, C. Dong, and Z. Zeng, “Toward sufficient spatial- frequency interaction for gradient-aware underwater image enhance- ment,” in ICASSP, 2024, pp. 3220–3224

2024
[11]

Twin adversarial contrastive learning for underwater image enhancement and beyond,

R. Liu, Z. Jiang, S. Yang, and X. Fan, “Twin adversarial contrastive learning for underwater image enhancement and beyond,” TIP, vol. 31, pp. 4922–4936, 2022

2022
[12]

A wavelet-based dual-stream network for underwater image enhancement,

Z. Ma and C. Oh, “A wavelet-based dual-stream network for underwater image enhancement,” in ICASSP, 2022, pp. 2769–2773

2022
[13]

Pugan: Physical model-guided underwater image enhance- ment using gan with dual-discriminators,

R. Cong, W. Yang, W. Zhang, C. Li, C.-L. Guo, Q. Huang, and S. Kwong, “Pugan: Physical model-guided underwater image enhance- ment using gan with dual-discriminators,” TIP, 2023

2023
[14]

Domain adaptation for underwater image enhancement,

Z. Wang, L. Shen, M. Xu, M. Yu, K. Wang, and Y . Lin, “Domain adaptation for underwater image enhancement,” TIP, vol. 32, pp. 1442– 1457, 2023

2023
[15]

U-shape transformer for underwater image enhancement,

L. Peng, C. Zhu, and L. Bian, “U-shape transformer for underwater image enhancement,” TIP, 2023

2023
[16]

Re- inforced swin-convs transformer for simultaneous underwater sensing scene image enhancement and super-resolution,

T. Ren, H. Xu, G. Jiang, M. Yu, X. Zhang, B. Wang, and T. Luo, “Re- inforced swin-convs transformer for simultaneous underwater sensing scene image enhancement and super-resolution,” IEEE TGRS , vol. 60, pp. 1–16, 2022

2022
[17]

Contrastive semi- supervised learning for underwater image restoration via reliable bank,

S. Huang, K. Wang, H. Liu, J. Chen, and Y . Li, “Contrastive semi- supervised learning for underwater image restoration via reliable bank,” in CVPR, 2023, pp. 18 145–18 155

2023
[18]

Unsu- pervised underwater image restoration: From a homology perspective,

Z. Fu, H. Lin, Y . Yang, S. Chai, L. Sun, Y . Huang, and X. Ding, “Unsu- pervised underwater image restoration: From a homology perspective,” in AAAI, vol. 36, 2022, pp. 643–651

2022
[19]

A generalized physical-knowledge-guided dynamic model for underwater image en- hancement,

P. Mu, H. Xu, Z. Liu, Z. Wang, S. Chan, and C. Bai, “A generalized physical-knowledge-guided dynamic model for underwater image en- hancement,” in ACM MM, 2023, pp. 7111–7120

2023
[20]

Test-time intensity consistency adaptation for shadow detection,

L. Zhu, W. Liu, X. Chen, Z. Li, X. Chen, Z. Wang, and C.-M. Pun, “Test-time intensity consistency adaptation for shadow detection,” arXiv, 2024

2024
[21]

Muraldiff: Diffusion for ancient murals restoration on large-scale pre-training,

Z. Xu, X. Zhang, W. Chen, J. Liu, T. Xu, and Z. Wang, “Muraldiff: Diffusion for ancient murals restoration on large-scale pre-training,” TETCI, 2024

2024
[22]

From redundancy to relevance: Enhancing explainability in multimodal large language models,

X. Zhang, C. Shen, X. Yuan, S. Yan, L. Xie, W. Wang, C. Gu, H. Tang, and J. Ye, “From redundancy to relevance: Enhancing explainability in multimodal large language models,” arXiv, 2024

2024
[23]

Cross-domain visual prompting with spatial proximity knowledge dis- tillation for histological image classification,

X. Li, G. Huang, L. Cheng, G. Zhong, W. Liu, X. Chen, and M. Cai, “Cross-domain visual prompting with spatial proximity knowledge dis- tillation for histological image classification,” Journal of Biomedical Informatics, vol. 158, p. 104728, 2024

2024
[24]

Residual feature- reutilization inception network,

Y . He, W. Song, L. Li, T. Zhan, and W. Jiao, “Residual feature- reutilization inception network,” PR, vol. 152, p. 110439, 2024

2024
[25]

Nndf: A new neural detection network for aspect-category sentiment analysis,

L. Li, Y . He, and L. Li, “Nndf: A new neural detection network for aspect-category sentiment analysis,” in KSEM, 2022, pp. 339–355

2022
[26]

Spa-former: An effective and lightweight transformer for image shadow removal,

X. Zhang, Y . Zhao, C. Gu, C. Lu, and S. Zhu, “Spa-former: An effective and lightweight transformer for image shadow removal,” in IJCNN, 2023, pp. 1–8

2023
[27]

Forgeryttt: Zero-shot image manipulation localization with test-time training,

W. Liu, X. Shen, C.-M. Pun, and X. Cun, “Forgeryttt: Zero-shot image manipulation localization with test-time training,” arXiv, 2024

2024
[28]

High-fidelity document stain removal via a large-scale real- world dataset and a memory-augmented transformer,

M. Li, H. Sun, Y . Lei, X. Zhang, Y . Dong, Y . Zhou, Z. Li, and X. Chen, “High-fidelity document stain removal via a large-scale real- world dataset and a memory-augmented transformer,” in WACV, 2024

2024
[29]

Seeing clearly by layer two: Enhancing attention heads to alleviate hallucination in lvlms,

X. Zhang, Y . Quan, C. Gu, C. Shen, X. Yuan, S. Yan, H. Cheng, K. Wu, and J. Ye, “Seeing clearly by layer two: Enhancing attention heads to alleviate hallucination in lvlms,” arXiv, 2024

2024
[30]

Dh-gan: Image manipulation localiza- tion via a dual homology-aware generative adversarial network,

W. Liu, X. Cun, and C.-M. Pun, “Dh-gan: Image manipulation localiza- tion via a dual homology-aware generative adversarial network,” PR, p. 110658, 2024

2024
[31]

High-resolution document shadow removal via a large-scale real-world dataset and a frequency- aware shadow erasing net,

Z. Li, X. Chen, C.-M. Pun, and X. Cun, “High-resolution document shadow removal via a large-scale real-world dataset and a frequency- aware shadow erasing net,” in ICCV, 2023, pp. 12 449–12 458

2023
[32]

Dual-hybrid attention network for specular highlight removal,

X. Guo, X. Chen, S. Luo, S. Wang, and C.-M. Pun, “Dual-hybrid attention network for specular highlight removal,” in ACM MM , 2024, pp. 10 173–10 181

2024
[33]

Explicit visual prompting for low-level structure segmentations,

W. Liu, X. Shen, C.-M. Pun, and X. Cun, “Explicit visual prompting for low-level structure segmentations,” in CVPR, 2023, pp. 19 434–19 445

2023
[34]

Enlighten-anything: When segment anything model meets low-light image enhancement,

X. Zhang, H. Tang, C. Gu, and S. Zhu, “Enlighten-anything: When segment anything model meets low-light image enhancement,” arXiv preprint arXiv:2306.10286, 2023

2023 arXiv
[35]

Uir- net: a simple and effective baseline for underwater image restoration and enhancement,

X. Mei, X. Ye, X. Zhang, Y . Liu, J. Wang, J. Hou, and X. Wang, “Uir- net: a simple and effective baseline for underwater image restoration and enhancement,” Remote Sensing , vol. 15, no. 1, p. 39, 2022

2022
[36]

Instance-adaptive zero-shot chain-of-thought prompting,

X. Yuan, C. Shen, S. Yan, X. Zhang, L. Xie, W. Wang, R. Guan, Y . Wang, and J. Ye, “Instance-adaptive zero-shot chain-of-thought prompting,” arXiv, 2024

2024
[37]

Dopra: Decoding over-accumulation penalization and re-allocation in specific weighting layer,

J. Wei and X. Zhang, “Dopra: Decoding over-accumulation penalization and re-allocation in specific weighting layer,” arXiv, 2024

2024
[38]

Coordfill: Efficient high-resolution image inpainting via parameterized coordinate querying,

W. Liu, X. Cun, C.-M. Pun, M. Xia, Y . Zhang, and J. Wang, “Coordfill: Efficient high-resolution image inpainting via parameterized coordinate querying,” in AAAI, vol. 37, no. 2, 2023, pp. 1746–1754

2023
[39]

Depth- aware test-time training for zero-shot video object segmentation,

W. Liu, X. Shen, H. Li, X. Bi, B. Liu, C.-M. Pun, and X. Cun, “Depth- aware test-time training for zero-shot video object segmentation,” in CVPR, 2024, pp. 19 218–19 227

2024
[40]

Generalized uncertainty- based evidential fusion with hybrid multi-head attention for weak- supervised temporal action localization,

Y . He, L. Li, T. Zhan, W. Jiao, and C.-M. Pun, “Generalized uncertainty- based evidential fusion with hybrid multi-head attention for weak- supervised temporal action localization,” in ICASSP, 2024, pp. 3855– 3859

2024
[41]

Visual attention network,

M.-H. Guo, C.-Z. Lu, Z.-N. Liu, M.-M. Cheng, and S.-M. Hu, “Visual attention network,” cvm, vol. 9, no. 4, pp. 733–752, 2023

2023
[42]

Scaling up your kernels to 31x31: Revisiting large kernel design in cnns,

X. Ding, X. Zhang, J. Han, and G. Ding, “Scaling up your kernels to 31x31: Revisiting large kernel design in cnns,” in CVPR, 2022, pp. 11 963–11 975

2022
[43]

Omni-kernel network for image restoration,

Y . Cui, W. Ren, and A. Knoll, “Omni-kernel network for image restoration,” in AAAI, vol. 38, no. 2, 2024, pp. 1426–1434

2024
[44]

Large separable kernel attention: Rethinking the large kernel attention design in cnn,

K. W. Lau, L.-M. Po, and Y . A. U. Rehman, “Large separable kernel attention: Rethinking the large kernel attention design in cnn,” ESWA, vol. 236, p. 121352, 2024

2024
[45]

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,” TIP, vol. 29, pp. 4376–4389, 2019

2019
[46]

Fast underwater image enhancement for improved visual perception,

M. J. Islam, Y . Xia, and J. Sattar, “Fast underwater image enhancement for improved visual perception,” IEEE RAL, vol. 5, no. 2, pp. 3227–3234, 2020

2020
[47]

Real-world underwater enhancement: Challenges, benchmarks, and solutions under natural light,

R. Liu, X. Fan, M. Zhu, M. Hou, and Z. Luo, “Real-world underwater enhancement: Challenges, benchmarks, and solutions under natural light,” TCSVT, vol. 30, no. 12, pp. 4861–4875, 2020

2020

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.

[1] [1]

A variational framework for underwater image dehazing and deblurring,

J. Xie, G. Hou, G. Wang, and Z. Pan, “A variational framework for underwater image dehazing and deblurring,” TCSVT, vol. 32, no. 6, pp. 3514–3526, 2021

work page 2021

[2] [2]

Under- water image enhancement via minimal color loss and locally adaptive contrast enhancement,

W. Zhang, P. Zhuang, H.-H. Sun, G. Li, S. Kwong, and C. Li, “Under- water image enhancement via minimal color loss and locally adaptive contrast enhancement,” TIP, vol. 31, pp. 3997–4010, 2022

work page 2022

[3] [3]

Rank-one prior: Real-time scene recovery,

J. Liu, R. W. Liu, J. Sun, and T. Zeng, “Rank-one prior: Real-time scene recovery,” TPAMI, 2022

work page 2022

[4] [4]

Underwater camera: Improving visual perception via adaptive dark pixel prior and color correction,

J. Zhou, Q. Liu, Q. Jiang, W. Ren, K.-M. Lam, and W. Zhang, “Underwater camera: Improving visual perception via adaptive dark pixel prior and color correction,” IJCV, pp. 1–19, 2023

work page 2023

[5] [5]

Underwater image enhancement via weighted wavelet visual perception fusion,

W. Zhang, L. Zhou, P. Zhuang, G. Li, X. Pan, W. Zhao, and C. Li, “Underwater image enhancement via weighted wavelet visual perception fusion,” TCSVT, 2023

work page 2023

[6] [6]

Sienet: Siamese expansion network for image extrapolation,

X. Zhang, F. Chen, C. Wang, M. Tao, and G.-P. Jiang, “Sienet: Siamese expansion network for image extrapolation,” SPL, vol. 27, pp. 1590– 1594, 2020

work page 2020

[7] [7]

Shadclips: When parameter-efficient fine-tuning with multimodal meets shadow removal,

X. Zhang, Z. Xu, H. Tang, C. Gu, S. Zhu, and X. Guan, “Shadclips: When parameter-efficient fine-tuning with multimodal meets shadow removal,” 2024

work page 2024

[8] [8]

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,” TIP, vol. 30, pp. 4985–5000, 2021

work page 2021