Pith. sign in

REVIEW 3 major objections 4 minor 239 references

Degradation-Guided Underwater Image Restoration with Task-Oriented Latent Control

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read The paper claims that underwater restoration improves when degradation cues guide feature adaptation while a learned latent code regulates skip connections, achieving the best PSNR and LPIPS on U90 and LSUI-400 with a 2.61M-parameter…

desk verdict A careful, transparently-reported underwater restoration paper whose main SOTA claim rests on mixed-protocol benchmarks and a small internal inconsistency; worth refereeing, but the authors should release code and confirm the ranking under a unified evaluation. read the letter →

arxiv 2608.08661 v1 pith:MLAPZVHA submitted 2026-08-09 cs.CV

Xu Zhang , Xuhui Cao , Kangzhe Yuan , Laibin Chang , Yichu Xu , Shi Chen , Huan Zhang , Yong Chen This is my paper · ORCID
classification cs.CV
keywords underwaterimagerestorationdegradation-guidedfeaturemodulationtask-orientedlatentcontrolskipconnectionregulationU-Netenhancementdeeplearningbenchmarks
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

Underwater images carry degradation information that behaves two ways: it signals what needs correcting, and it contaminates the features a decoder reuses. PROTEUS is built on the claim that a restoration network should exploit both. Its degradation-guided feature modulation block steers feature processing with spatially varying cues, while a task-oriented latent controller learns a bottleneck code that decides which skip-connection channels to attenuate during decoding. On paired benchmarks the paper reports the best PSNR and LPIPS on U90 (25.50 dB and 0.081) and LSUI-400 (28.99 dB and 0.083), with a small footprint of 2.61M parameters and 18.52G FLOPs.

What carries the argument

The central machinery is the pair formed by the Guided Dynamic Feature Modulation Block (GDFMB) and the task-oriented latent controller. GDFMB uses a guide image derived from a Gray-Edge colour-bias prior, samples spatially varying degradation cues through deformable cross-attention, expands them through Gaussian basis functions with a selective fusion gate, and adds a Gradient Fusion Block to preserve edges. The latent controller maps the bottleneck feature into a control code $F_z$ and a complementary residual $F_{res}$, regularises $F_z$ with reference alignment, orthogonality, and contrastive loss, and uses it to generate channel-wise Attention Gate weights that rescale skip features before decoder fusion. Together they turn degradation information into two separated functions: guiding where features are transformed and deciding which skip channels to pass through, without assuming either representation is a physically clean image.

What would settle it

Retrain every compared method and PROTEUS from scratch in one codebase with identical UIEB splits, patch size, augmentation, and training epochs, then recompute U90 and LSUI-400 PSNR, SSIM, and LPIPS: if PROTEUS no longer holds the best PSNR and LPIPS, the claimed advantage is a setup artefact.

Watch

Extended reading notes

Core claim

PROTEUS claims that underwater degradation information should be treated as a dual-use signal rather than something to simply estimate and remove. Spatially varying degradation cues from the input are used to guide adaptive feature transformation at every encoder–decoder stage, while degradation-entangled encoder features are regulated before reuse by a bottleneck control code applied channel-wise to skip connections. The control code is explicitly task-oriented, not a clean-image embedding: it is regularised by alignment, orthogonality, and contrastive separation, and diagnostics reported by the paper show it is actually farther from the ground-truth latent than the unfiltered bottleneck feature. By coupling these two controls, the paper obtains best PSNR and LPIPS on U90 and LSUI-400, joint-best SSIM on U90, and competitive results across four non-reference benchmarks, at 2.61M parameters and 18.52G FLOPs.

Load-bearing premise

The headline comparison assumes that published benchmark numbers for competing methods can be fairly compared with PROTEUS numbers produced from the authors' own checkpoints; the paper explicitly calls these benchmark-level comparisons rather than jointly retrained results under a unified codebase, so if training protocols, data splits, or evaluation settings differ across methods, the reported gaps may be setup differences rather than architectural superiority.

Editorial extensions

If this is right

  • If PROTEUS is right, skip connections in underwater restoration networks should be seen as a regulated information pathway rather than a pass-through of encoder features.
  • The two controls are complementary: ablations show that removing either the latent-conditioned Attention Gate or the deformable spatial guidance lowers PSNR, and visual analyses show combining them gives the best colour and detail.
  • The reported representation diagnostics indicate that a useful control code need not be closer to a clean-image latent, which supports task-oriented regularisation for restoration pipelines that cannot assume a clean/degraded decomposition.
  • The claimed gains come with a small footprint of 2.61M parameters and 18.52G FLOPs, so the dual-control strategy is a plausible candidate for lightweight underwater imaging systems.

Reading between the lines

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

  • Because the paper finds the raw underwater input nearly matches the preprocessed Gray-Edge guide in guiding the network, a testable extension is to replace the hand-designed guide with a learned adaptable preprocessing module and measure whether spatial guidance gains grow.
  • The task-oriented control code, which is explicitly not a clean-latent projection, may transfer to other ill-posed restoration settings such as low-light or scattering-medium imaging where clean/degraded separation is not well defined.
  • The gate diagnostics showing stable channel modulation with weak per-channel degradation correlation suggest the controller could be exposed as a user-adjustable colour and contrast control in practical underwater systems, a role the paper does not claim.
  • Per-image and cross-dataset interventions show spatial guidance helps most but not all images, so a confidence or severity estimator that decides when to disable spatial guidance is a natural follow-up.
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. PROTEUS is a U-Net-based underwater image restoration method that couples two mechanisms: a Guided Dynamic Feature Modulation Block (GDFMB), which uses a Gray-Edge-derived guide image and deformable cross-attention with Gaussian basis expansion to adapt feature processing across stages, and a task-oriented latent controller that maps the bottleneck feature into a control code F_z, regularizes it with alignment, orthogonality, and contrastive losses, and uses it to modulate skip-feature channels through Attention Gates. The paper claims highly competitive restoration performance on five paired benchmarks (U90, LSUI, UFO-120, EUVP-Scene, EUVP-Dark) and four non-reference benchmarks, with a favorable cost profile of 2.61M parameters and 18.52G FLOPs. The evidence includes ablations, same-checkpoint test-time interventions, bootstrap confidence intervals, cross-dataset transfer tests, and explicit representation diagnostics that avoid overclaiming a clean-latent interpretation of F_z.

Significance. If the headline results hold under a unified evaluation protocol, PROTEUS would offer a strong quality/cost trade-off and a reusable idea: using a structured bottleneck code to regulate skip-feature reuse rather than merely to condition the decoder. The manuscript is more careful than many papers in this area: the same-checkpoint interventions with paired bootstrap CIs, per-image gain distributions, real-degradation stratification, and the explicit statement that F_z is not a metrically cleaner embedding all provide genuinely useful evidence for the architectural claims. However, the central quantitative claim of state-of-the-art ranking currently depends on a benchmark-level comparison that mixes previously published numbers with the authors' own checkpoints, and there is an internal inconsistency between the main tables and the supplementary tables. These issues are fixable but must be addressed before the ranking claims can be considered established.

major comments (3)
  1. [Supplementary, 'Evaluation protocol and result provenance'; Tables 1–2] The headline claim that PROTEUS achieves the best PSNR and LPIPS on U90 and LSUI-400 is not yet supported by a controlled comparison. The supplement states that the benchmark tables 'combine results reported by the cited methods with PROTEUS results evaluated using our checkpoints' and should be viewed as 'benchmark-level comparisons rather than jointly retrained results under a unified codebase.' Given the small margins (0.39 dB over UniUIR on U90 and 0.57 dB over UniUIR on LSUI-400), differences in input resizing, test-crop strategy, LPIPS backbone version, or test split can plausibly change the ranking. The authors should either re-evaluate the closest baselines under the same protocol used for PROTEUS (resize to 256×256, AlexNet LPIPS, identical split) or explicitly reframe the claims as benchmark-level, with all ranking statements made conditional on that caveat.
  2. [Table 1 vs. Table S3] There is an internal inconsistency in the reported full-model metrics. Main Table 1 lists PROTEUS on U90 as PSNR 25.50, SSIM 0.933, LPIPS 0.081 and on LSUI-400 as PSNR 28.99, SSIM 0.913, LPIPS 0.083. Table S3, described as the same checkpoint evaluated on all 90 U90 pairs and 400 LSUI pairs, gives U90 SSIM 0.935 and LSUI SSIM 0.921 for the 'Full spatial guide' row, with identical PSNR and LPIPS values. Since the PSNR and LPIPS match exactly, simple rounding cannot explain both SSIM discrepancies (0.933 vs. 0.935 and 0.913 vs. 0.921). The authors must reconcile these numbers, state explicitly how unrounded values are handled, and ensure that every table reporting the same model reports the same metric values, otherwise the reliability of the quantitative results cannot be audited.
  3. [Supplementary, Table S5 and 'Latent-Loss Weight Sensitivity'] The paper should more prominently flag that the latent-loss weight sensitivity study is a fixed-coordinate local adaptation experiment, not an ablation of the trained model. The supplement correctly says 'adapted scores are not comparable to the unadapted 25.50 dB main-paper result,' but the main text's representation-and-gate diagnostics paragraph cites the three-seed analysis without this caveat. This matters because the same section reports that increasing the alignment weight reduces but never eliminates the F_z-vs-raw-bottleneck deficit, which is an important negative result for the clean-latent interpretation. The main text should carry the same qualification so readers do not mistake the local-sensitivity numbers for main-model properties.
minor comments (4)
  1. [Supplementary, Table S2 and 'Additional Ablation Results'] The loss-removal rows in Table S2 are independently trained runs, not seed-paired interventions, and the supplement already describes them as 'descriptive ablation evidence rather than paired statistical effects.' The main text should repeat this caveat in the paragraph that cites the loss ablations, rather than presenting the 25.50-to-25.24 drop as a precisely measured effect.
  2. [References] The reference list contains duplicated entries: Islam et al. 2020a and 2020b refer to the same paper, and Zhang et al. 2022a and 2022b also appear to be the same publication. These duplicates should be merged and cited with distinct keys only where different works are intended.
  3. [Figure 2] Figure 2 is dense and has a very long caption; in the final version the module names, data-flow arrows, and augmentation branches should be redrawn at larger size and with legible labels, because several components (GDAB, GBE, Attention Gate, photometric branch) are difficult to resolve from the current rendering.
  4. [Tables 1–2, FLOPs reporting] The FLOPs and parameter counts for competing methods are taken from their respective papers, and the supplement does not state whether the same FLOPs-counting conventions (e.g., whether batch dimensions or the final convolution are included) apply to all entries. A short footnote stating the counting tool and input size for all rows would prevent apples-to-oranges efficiency comparisons.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular derivation: both claimed controls are validated by ablations and honest diagnostics; benchmark-table mixing is a fairness risk, not circularity.

full rationale

The paper's derivation chain is self-contained. PROTEUS is trained end-to-end with image-level losses, and the two proposed mechanisms are validated by same-checkpoint ablations and interventions rather than by definition. The main potential circularity—that the 'task-oriented' code Fz is just a relabeled clean-latent projection—is explicitly tested and rejected: the paper reports that Fz is farther from the ground-truth latent than the raw bottleneck (MSE 0.0164 vs 0.0145; cosine 0.7162 vs 0.8067), and the supplement states that no relative-weight setting crosses zero on these diagnostics. The reference-latent loss Lalign is training supervision, not a predicted quantity, so it is not a fitted input renamed as a prediction. The skip gate is a standard channel-wise modulation, but the paper ablates against CBAM and shows a difference, so it is not a pure renaming of a known block. The only load-adjacent weakness is benchmark provenance: the supplement warns that the tables are 'benchmark-level comparisons rather than jointly retrained results under a unified codebase,' and main Table 1 reports U90 SSIM 0.933 while Table S3 reports 0.935 for the same checkpoint and test set; these are correctness/fairness issues, not circularity. Self-citations (UniUIR, CDF-UIE, etc.) appear in related work and baselines but are not used to justify the method's premises. No uniqueness theorem or ansatz is imported from the authors' prior work. Hence no circular step; score 2 only for minor, non-load-bearing self-citations.

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

The central claims rest on standard training hyperparameters and domain assumptions about underwater image formation and reference data. No new physical entities are introduced. The most consequential assumptions are the validity of the paired reference datasets and the Gray-Edge prior as a degradation cue.

free parameters (5)
  • K (number of Gaussian basis functions) = not specified
    In Equation (7), the Gaussian expansion uses K basis functions. Chosen by hand, not fitted to data, though performance may depend on it.
  • offset range r = not specified
    In Equation (6), r controls the range of deformable offsets in GDAB. A hyperparameter that affects spatial guidance strength.
  • residual blend ratio = 0.5
    Residual blend ratio in the GDFMB; selected by hand to balance gradient fusion.
  • latent loss weights (lambda_a, lambda_o, lambda_c) = (0.1, 0.05, 0.05)
    Weights in Equation (15) are set to these values. Table S5 shows sensitivity but no optimality criterion.
  • beta0 (self-supervised consistency weight) = 0.1
    Initial weight of the self-supervised consistency loss in Equation (17), decaying to zero.
assumptions (4)
  • domain assumption The Gray-Edge prior holds approximately for underwater scenes: the average edge magnitude is achromatic across channels.
    Used in the Gray-Edge Prior Module (Equation 4). Violations could bias the guide image and degrade conditioning.
  • domain assumption The global residual formulation I_r = D(...) + I_d is sufficient for restoration.
    Assumes the network is trained to predict the residual. If this assumption fails, output may not converge to the desired clean image.
  • domain assumption Supervision from paired underwater and reference images provides a valid training signal for natural appearance.
    All paired benchmarks rely on reference images that may not fully represent real underwater degradation or may encode a specific color balance.
  • standard math Standard stochastic gradient optimization with the composite loss converges to a useful local optimum.
    Training relies on empirical risk minimization; no convergence guarantee is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Degradation-Guided Underwater Image Restoration with Task-Oriented Latent Control." pith.science (2026). https://pith.science/paper/MLAPZVHA

@misc{pith2026260808661,
  author       = {Pith},
  title        = {Pith review of: Degradation-Guided Underwater Image Restoration with Task-Oriented Latent Control},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MLAPZVHA}},
  note         = {Machine review of arXiv:2608.08661}
}
read the original abstract

Degradation information in underwater images plays a dual role: its spatial and spectral cues can guide adaptive restoration, while degradation-entangled features may be propagated without explicit regulation during decoding. Existing methods largely overlook this dual role, either underexploiting degradation cues or directly forwarding encoder features through skip connections. To address this issue, we propose PROTEUS, which couples degradation-guided feature adaptation with task?oriented latent control. PROTEUS tackles this problem from two complementary perspectives. At the feature level, the Guided Dynamic Feature Modulation Block exploits spatially varying degradation cues to adapt feature processing across network stages. At the representation level, the task-oriented latent controller learns a structured control code under discriminative regularisation and uses it for channel-wise modulation of skip features, without requiring the code to form a metrically cleaner embedding. Extensive experiments on five paired and four non-reference underwater benchmarks demonstrate that PROTEUS achieves highly competitive restoration performance, with a favourable balance between restoration quality and computational cost.

Figures

Figures reproduced from arXiv: 2608.08661 by the authors.

Figure 1
Figure 1. Motivation and overview of PROTEUS. (a) Degra [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Proposed network architecture of PROTEUS. The framework combines degradation-guided feature processing and [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Full visual comparison on U90 (Li et al. 2020) dataset. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Visual analysis on U90 dataset. “Uniform guide” [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

239 extracted references · 75 canonical work pages

  1. [1]

    ECCV , pages=

    Onerestore: A universal restoration framework for composite degradation , author=. ECCV , pages=

  2. [2]

    AAAI , pages=

    Waterdiffusion: Learning a prior-involved unrolling diffusion for joint underwater saliency detection and visual restoration , author=. AAAI , pages=

  3. [3]

    ACM MM , pages=

    UIS-Mamba: exploring mamba for underwater instance segmentation via dynamic tree scan and hidden state weaken , author=. ACM MM , pages=

  4. [4]

    AAAI , pages=

    Empowering dino representations for underwater instance segmentation via aligner and prompter , author=. AAAI , pages=

  5. [5]

    TCSVT , volume=

    Underwater Scene Clarity Reconstruction via Multilayer Information Fusion and Self-Organized Stitching , author=. TCSVT , volume=

  6. [6]

    ICRA , pages=

    End-to-end underwater multi-view stereo for dense scene reconstruction , author=. ICRA , pages=

  7. [7]

    Cong, Runmin and Zhang, Yumo and Fang, Leyuan and Li, Jun and Zhao, Yao and Kwong, Sam , journal=

  8. [8]

    TGRS , volume=

    Nested network with two-stream pyramid for salient object detection in optical remote sensing images , author=. TGRS , volume=

Show all 239 references
  1. [9]

    AAAI , volume=

    Unsupervised underwater image restoration: From a homology perspective , author=. AAAI , volume=

  2. [10]

    TIP , volume=

    Edge-based color constancy , author=. TIP , volume=

  3. [11]

    CVPR , pages=

    Hvi: A new color space for low-light image enhancement , author=. CVPR , pages=

  4. [12]

    TIP , volume=

    Image quality assessment: from error visibility to structural similarity , author=. TIP , volume=

  5. [13]

    arXiv preprint arXiv:2505.21637 , year=

    BaryIR: Learning multi-source unified representation in continuous barycenter space for generalizable all-in-one image restoration , author=. arXiv preprint arXiv:2505.21637 , year=

    arXiv

  6. [14]

    JOE , volume=

    Computer modeling and the design of optimal underwater imaging systems , author=. JOE , volume=

  7. [15]

    TPAMI , volume=

    Single image haze removal using dark channel prior , author=. TPAMI , volume=

  8. [16]

    CG&A , volume=

    Underwater depth estimation and image restoration based on single images , author=. CG&A , volume=

  9. [17]

    TIP , volume=

    Underwater image restoration based on image blurriness and light absorption , author=. TIP , volume=

  10. [18]

    TPAMI , volume=

    Underwater single image color restoration using haze-lines and a new quantitative dataset , author=. TPAMI , volume=

  11. [19]

    Graphics Gems IV , pages=

    Contrast limited adaptive histogram equalization , author=. Graphics Gems IV , pages=. 1994 , publisher=

    1994

  12. [20]

    CVPR , pages=

    Enhancing underwater images and videos by fusion , author=. CVPR , pages=

  13. [21]

    ICIP , pages=

    A retinex-based enhancing approach for single underwater image , author=. ICIP , pages=

  14. [22]

    TIP , volume=

    Underwater image enhancement via minimal color loss and locally adaptive contrast enhancement , author=. TIP , volume=

  15. [23]

    TIP , volume=

    An underwater image enhancement benchmark dataset and beyond , author=. TIP , volume=

  16. [24]

    ICRA , pages=

    Enhancing underwater imagery using generative adversarial networks , author=. ICRA , pages=

  17. [25]

    RA-L , volume=

    Fast underwater image enhancement for improved visual perception , author=. RA-L , volume=

  18. [26]

    PR , volume=

    Underwater scene prior inspired deep underwater image and video enhancement , author=. PR , volume=

  19. [27]

    AAAI , volume=

    Shallow-UWnet: Compressed model for underwater image enhancement , author=. AAAI , volume=

  20. [28]

    TIP , volume=

    Underwater image enhancement via medium transmission-guided multi-color space embedding , author=. TIP , volume=

  21. [29]

    ECCV , pages=

    Uncertainty inspired underwater image enhancement , author=. ECCV , pages=

  22. [30]

    TIP , volume=

    U-Shape Transformer for Underwater Image Enhancement , author=. TIP , volume=

  23. [31]

    BMVC , year=

    Five A+ Network: You Only Need 9K Parameters for Underwater Image Enhancement , author=. BMVC , year=

  24. [32]

    CVPR , pages=

    Wavelet-Based Fourier Information Interaction with Frequency Diffusion Adjustment for Underwater Image Restoration , author=. CVPR , pages=

  25. [33]

    IJCNN , pages=

    UWFormer: Underwater Image Enhancement via a Semi-Supervised Multi-Scale Transformer , author=. IJCNN , pages=

  26. [34]

    CVPR , pages=

    Contrastive semi-supervised learning for underwater image restoration via reliable bank , author=. CVPR , pages=

  27. [35]

    arXiv preprint , year=

    BaryIR: Barycentric Coordinates-Inspired Latent Purification for Image Restoration , author=. arXiv preprint , year=

  28. [36]

    CVPR , pages=

    Squeeze-and-excitation networks , author=. CVPR , pages=

  29. [37]

    ECCV , pages=

    CBAM: Convolutional block attention module , author=. ECCV , pages=

  30. [38]

    ICLR , year=

    An image is worth 16x16 words: Transformers for image recognition at scale , author=. ICLR , year=

  31. [39]

    ICCVW , pages=

    SwinIR: Image restoration using swin transformer , author=. ICCVW , pages=

  32. [40]

    CVPR , pages=

    Restormer: Efficient transformer for high-resolution image restoration , author=. CVPR , pages=

  33. [41]

    Deformable

    Zhu, Xizhou and Su, Weijie and Lu, Lewei and Li, Bin and Wang, Xiaogang and Dai, Jifeng , booktitle=. Deformable

  34. [42]

    CVPR , pages=

    Vision transformer with deformable attention , author=. CVPR , pages=

  35. [43]

    ICCV , pages=

    Dual Aggregation Transformer for Image Super-Resolution , author=. ICCV , pages=

  36. [44]

    CVPR , pages=

    Learning a sparse transformer network for effective image deraining , author=. CVPR , pages=

  37. [45]

    IEEE RAM , volume=

    WaterFormer: A Global--Local Transformer for Underwater Image Enhancement with Environment Adaptor , author=. IEEE RAM , volume=. 2024 , doi=

    2024

  38. [46]

    CVPR , pages=

    Recovering realistic texture in image super-resolution by deep spatial feature transform , author=. CVPR , pages=

  39. [47]

    ICCV , pages=

    Arbitrary style transfer in real-time with adaptive instance normalization , author=. ICCV , pages=

  40. [48]

    CVPR , pages=

    Unsupervised degradation representation learning for blind super-resolution , author=. CVPR , pages=

  41. [49]

    CVPR , pages=

    All-in-one image restoration for unknown corruption , author=. CVPR , pages=

  42. [50]

    NeurIPS , pages=

    PromptIR: Prompting for All-in-One Blind Image Restoration , author=. NeurIPS , pages=

  43. [51]

    CVPR , pages=

    Ingredient-oriented multi-degradation learning for image restoration , author=. CVPR , pages=

  44. [52]

    CVPR , pages=

    Momentum contrast for unsupervised visual representation learning , author=. CVPR , pages=

  45. [53]

    ICML , pages=

    A simple framework for contrastive learning of visual representations , author=. ICML , pages=

  46. [54]

    CVPR , pages=

    Contrastive learning for compact single image dehazing , author=. CVPR , pages=

  47. [55]

    IEEE Access , volume=

    Domain Adaptation for Underwater Image Enhancement via Content and Style Separation , author=. IEEE Access , volume=

  48. [56]

    CVPR , pages=

    HVI: A New Color Space for Low-Light Image Enhancement , author=. CVPR , pages=

  49. [57]

    TIP , volume=

    Zhang, Xu and Zhang, Huan and Wang, Guoli and Zhang, Qian and Zhang, Lefei and Du, Bo , title=. TIP , volume=

  50. [58]

    CVPR , pages=

    Complexity experts are task-discriminative learners for any image restoration , author=. CVPR , pages=

  51. [59]

    RSS , year=

    Islam, Md Jahidul and Luo, Peigen and Sattar, Junaed , title=. RSS , year=

  52. [60]

    WACV , pages=

    Phaseformer: Phase-based attention mechanism for underwater image restoration and beyond , author=. WACV , pages=

  53. [61]

    AAAI , pages=

    Synergistic multiscale detail refinement via intrinsic supervision for underwater image enhancement , author=. AAAI , pages=

  54. [62]

    TGRS , volume=

    CDF-UIE: Leveraging Cross-Domain Fusion for Underwater Image Enhancement , author=. TGRS , volume=

  55. [63]

    WACV , pages=

    WWE-UIE: A wavelet & white balance efficient network for underwater image enhancement , author=. WACV , pages=

  56. [64]

    TMM , volume=

    UIQI: A comprehensive quality evaluation index for underwater images , author=. TMM , volume=

  57. [65]

    UIF: An Objective Quality Assessment for Underwater Image Enhancement , year=

    Zheng, Yannan and Chen, Weiling and Lin, Rongfu and Zhao, Tiesong and Le Callet, Patrick , journal=. UIF: An Objective Quality Assessment for Underwater Image Enhancement , year=

  58. [66]

    ESWA , volume=

    Enhancing 3D video watching experiences: Tackling compression and 3D warping distortions in synthesized view with perceptual guidance , author=. ESWA , volume=. 2025 , publisher=

    2025

  59. [67]

    PR , pages=

    Unveiling the underwater world: CLIP perception model-guided underwater image enhancement , author=. PR , pages=. 2025 , publisher=

    2025

  60. [68]

    ICLR , year=

    AdaIR: Adaptive All-in-One Image Restoration via Frequency Mining and Modulation , author=. ICLR , year=

  61. [69]

    NeurIPS , pages=

    Gans trained by a two time-scale update rule converge to a local nash equilibrium , author=. NeurIPS , pages=

  62. [70]

    Non-Uniform Illumination Underwater Image Restoration via Illumination Channel Sparsity Prior , year=

    Hou, Guojia and Li, Nan and Zhuang, Peixian and Li, Kunqian and Sun, Haihan and Li, Chongyi , journal=. Non-Uniform Illumination Underwater Image Restoration via Illumination Channel Sparsity Prior , year=

  63. [71]

    TCSVT , volume=

    Fan, Junyu and Xu, Jie and Zhou, Jingchun and Meng, Danling and Lin, Yi , title=. TCSVT , volume=

  64. [72]

    UIE-FSMC: Underwater Image Enhancement Based on Few-Shot Learning and Multi-Color Space , year=

    Qiao, Nianzu and Sun, Jia and Ge, Quanbo and Sun, Changyin , journal=. UIE-FSMC: Underwater Image Enhancement Based on Few-Shot Learning and Multi-Color Space , year=

  65. [73]

    Underwater Image Enhancement via Weighted Wavelet Visual Perception Fusion , year=

    Zhang, Weidong and Zhou, Ling and Zhuang, Peixian and Li, Guohou and Pan, Xipeng and Zhao, Wenyi and Li, Chongyi , journal=. Underwater Image Enhancement via Weighted Wavelet Visual Perception Fusion , year=

  66. [74]

    Cross-Scope Spatial-Spectral Information Aggregation for Hyperspectral Image Super-Resolution , year=

    Chen, Shi and Zhang, Lefei and Zhang, Liangpei , journal=. Cross-Scope Spatial-Spectral Information Aggregation for Hyperspectral Image Super-Resolution , year=

  67. [75]

    PR , volume=

    Two-step domain adaptation for underwater image enhancement , author=. PR , volume=. 2022 , publisher=

    2022

  68. [76]

    S³ANet: Spatial–Spectral Self-Attention Learning Network for Defending Against Adversarial Attacks in Hyperspectral Image Classification , year=

    Xu, Yichu and Xu, Yonghao and Jiao, Hongzan and Gao, Zhi and Zhang, Lefei , journal=. S³ANet: Spatial–Spectral Self-Attention Learning Network for Defending Against Adversarial Attacks in Hyperspectral Image Classification , year=

  69. [77]

    arXiv preprint arXiv:2409.07723 , year=

    Advancing Depth Anything Model for Unsupervised Monocular Depth Estimation in Endoscopy , author=. arXiv preprint arXiv:2409.07723 , year=

    arXiv

  70. [78]

    CVPR , pages=

    OSMamba: Omnidirectional Spectral Mamba with Dual-Domain Prior Generator for Exposure Correction , author=. CVPR , pages=

  71. [79]

    ICML , pages=

    See more details: Efficient image super-resolution by experts mining , author=. ICML , pages=

  72. [80]

    COLM , year=

    Mamba: Linear-Time Sequence Modeling with Selective State Spaces , author=. COLM , year=

  73. [81]

    CVPR , pages=

    Multimodal Prompt Perceiver: Empower Adaptiveness Generalizability and Fidelity for All-in-One Image Restoration , author=. CVPR , pages=

  74. [82]

    ACM MM , pages=

    Underwater image enhancement by transformer-based diffusion model with non-uniform sampling for skip strategy , author=. ACM MM , pages=

  75. [83]

    ACM MM , pages=

    A generalized physical-knowledge-guided dynamic model for underwater image enhancement , author=. ACM MM , pages=

  76. [84]

    ICCV , pages=

    Segment anything , author=. ICCV , pages=

  77. [85]

    Depth Anything V2 , year =

    Yang, Lihe and Kang, Bingyi and Huang, Zilong and Zhao, Zhen and Xu, Xiaogang and Feng, Jiashi and Zhao, Hengshuang , booktitle=. Depth Anything V2 , year =

  78. [86]

    CVIU , volume=

    BacklitNet: A dataset and network for backlit image enhancement , author=. CVIU , volume=. 2022 , publisher=

    2022

  79. [87]

    CVPR , pages=

    SNR-aware low-light image enhancement , author=. CVPR , pages=

  80. [88]

    ICCV , pages=

    Diff-retinex: Rethinking low-light image enhancement with a generative diffusion model , author=. ICCV , pages=

  81. [89]

    ICLR , year=

    Denoising Diffusion Implicit Models , author=. ICLR , year=

  82. [90]

    NeurIPS , pages=

    Denoising diffusion probabilistic models , author=. NeurIPS , pages=

  83. [91]

    Neurocomputing , volume=

    Srdiff: Single image super-resolution with diffusion probabilistic models , author=. Neurocomputing , volume=. 2022 , publisher=

    2022

  84. [92]

    IJCV , pages=

    Exploiting diffusion prior for real-world image super-resolution , author=. IJCV , pages=. 2024 , publisher=

    2024

  85. [93]

    TOG , volume=

    Low-light image enhancement with wavelet-based diffusion models , author=. TOG , volume=. 2023 , publisher=

    2023

  86. [94]

    CVPR , pages=

    Light the Night: A Multi-Condition Diffusion Framework for Unpaired Low-Light Enhancement in Autonomous Driving , author=. CVPR , pages=

  87. [95]

    CVPR , pages=

    Wavelet-based fourier information interaction with frequency diffusion adjustment for underwater image restoration , author=. CVPR , pages=

  88. [96]

    ESWA , volume=

    UIEDP: Boosting underwater image enhancement with diffusion prior , author=. ESWA , volume=. 2025 , publisher=

    2025

  89. [97]

    Underwater Image Enhancement via Principal Component Fusion of Foreground and Background , year=

    Zhang, Weidong and Liu, Qingmin and Feng, Yikun and Cai, Lei and Zhuang, Peixian , journal=. Underwater Image Enhancement via Principal Component Fusion of Foreground and Background , year=

  90. [98]

    CVIU , volume=

    Joint Multi-Dimensional Dynamic Attention and Transformer for General Image Restoration , author=. CVIU , volume=

  91. [99]

    TCSVT , volume=

    FDCE-Net: Underwater Image Enhancement with Embedding Frequency and Dual Color Encoder , author=. TCSVT , volume=

  92. [100]

    IJCV , volume=

    HCLR-net: Hybrid contrastive learning regularization with locally randomized perturbation for underwater image enhancement , author=. IJCV , volume=

  93. [101]

    IJCV , pages=

    Underwater image restoration via adaptive dark pixel prior and color correction , author=. IJCV , pages=

  94. [102]

    Perceive-IR: Learning to Perceive Degradation Better for All-in-One Image Restoration , year=

    Zhang, Xu and Ma, Jiaqi and Wang, Guoli and Zhang, Qian and Zhang, Huan and Zhang, Lefei , journal=. Perceive-IR: Learning to Perceive Degradation Better for All-in-One Image Restoration , year=

  95. [103]

    TIP , volume=

    Underwater image enhancement with hyper-laplacian reflectance priors , author=. TIP , volume=. 2022 , publisher=

    2022

  96. [104]

    arXiv preprint arXiv:2405.08419 , year=

    WaterMamba: Visual State Space Model for Underwater Image Enhancement , author=. arXiv preprint arXiv:2405.08419 , year=

    arXiv

  97. [105]

    DiffWater: Underwater Image Enhancement Based on Conditional Denoising Diffusion Probabilistic Model , year=

    Guan, Meisheng and Xu, Haiyong and Jiang, Gangyi and Yu, Mei and Chen, Yeyao and Luo, Ting and Zhang, Xuebo , journal=. DiffWater: Underwater Image Enhancement Based on Conditional Denoising Diffusion Probabilistic Model , year=

  98. [106]

    RA-L , volume=

    Fast underwater image enhancement for improved visual perception , author=. RA-L , volume=. 2020 , publisher=

    2020

  99. [107]

    TCSVT , volume=

    Real-world underwater enhancement: Challenges, benchmarks, and solutions under natural light , author=. TCSVT , volume=. 2020 , publisher=

    2020

  100. [108]

    IEEE Access , volume=

    An in-depth survey of underwater image enhancement and restoration , author=. IEEE Access , volume=. 2019 , publisher=

    2019

  101. [109]

    IJCA , volume=

    A Survey on Underwater Image Enhancement Techniques , author=. IJCA , volume=

  102. [110]

    2011 , publisher=

    Marine ecology: processes, systems, and impacts , author=. 2011 , publisher=

    2011

  103. [111]

    JENRL , volume=

    The Marine Strategy Framework Directive: a new European approach to the regulation of the marine environment, marine natural resources and marine ecological services , author=. JENRL , volume=. 2011 , publisher=

    2011

  104. [112]

    Oikos , volume=

    Towards an urban marine ecology: characterizing the drivers, patterns and processes of marine ecosystems in coastal cities , author=. Oikos , volume=. 2019 , publisher=

    2019

  105. [113]

    EURASIP JASP , volume=

    Underwater image processing: state of the art of restoration and image enhancement methods , author=. EURASIP JASP , volume=. 2010 , publisher=

    2010

  106. [114]

    SMC , pages=

    Enhancing the low quality images using unsupervised colour correction method , author=. SMC , pages=

  107. [115]

    TIP , volume=

    Underwater image enhancement by dehazing with minimum information loss and histogram distribution prior , author=. TIP , volume=. 2016 , publisher=

    2016

  108. [116]

    Object Detection in 20 Years: A Survey , author=. Proc. IEEE , volume=

  109. [117]

    JOE , volume=

    Underwater image enhancement using a multiscale dense generative adversarial network , author=. JOE , volume=. 2019 , publisher=

    2019

  110. [118]

    2017 , publisher=

    Li, Jie and Skinner, Katherine A and Eustice, Ryan M and Johnson-Roberson, Matthew , journal=. 2017 , publisher=

    2017

  111. [119]

    TIP , volume=

    Underwater image enhancement via medium transmission-guided multi-color space embedding , author=. TIP , volume=. 2021 , publisher=

    2021

  112. [120]

    CVPR , pages=

    Sea-thru: A method for removing water from underwater images , author=. CVPR , pages=

  113. [121]

    TIP , volume=

    Peng, Lintao and Zhu, Chunli and Bian, Liheng , title=. TIP , volume=

  114. [122]

    AAAI , pages=

    Underwater ranker: Learn which is better and how to be better , author=. AAAI , pages=

  115. [123]

    AAAI , pages=

    Exploring clip for assessing the look and feel of images , author=. AAAI , pages=

  116. [124]

    CVPR , pages=

    Curricular contrastive regularization for physics-aware single image dehazing , author=. CVPR , pages=

  117. [125]

    ICML , pages=

    Learning transferable visual models from natural language supervision , author=. ICML , pages=. 2021 , organization=

    2021

  118. [126]

    TIP , volume=

    Generalization of the dark channel prior for single image restoration , author=. TIP , volume=. 2018 , publisher=

    2018

  119. [127]

    JVCIR , volume=

    Automatic red-channel underwater image restoration , author=. JVCIR , volume=. 2015 , publisher=

    2015

  120. [128]

    TIP , volume=

    Underwater image restoration based on image blurriness and light absorption , author=. TIP , volume=. 2017 , publisher=

    2017

  121. [129]

    AAAI , pages=

    Shallow-uwnet: Compressed model for underwater image enhancement (student abstract) , author=. AAAI , pages=

  122. [130]

    2021 , publisher=

    Wang, Yudong and Guo, Jichang and Gao, Huan and Yue, Huihui , journal=. 2021 , publisher=

    2021

  123. [131]

    ECCV , pages=

    Open-vocabulary detr with conditional matching , author=. ECCV , pages=. 2022 , organization=

    2022

  124. [132]

    ICLR , year=

    F-VLM: Open-Vocabulary Object Detection upon Frozen Vision and Language Models , author=. ICLR , year=

  125. [133]

    ECCV , pages=

    Extract free dense labels from clip , author=. ECCV , pages=. 2022 , organization=

    2022

  126. [134]

    CVPR , pages=

    Blind image quality assessment via vision-language correspondence: A multitask learning perspective , author=. CVPR , pages=

  127. [135]

    Ke, Junjie and Ye, Keren and Yu, Jiahui and Wu, Yonghui and Milanfar, Peyman and Yang, Feng , booktitle=

  128. [136]

    SPIC , volume=

    A reference-free underwater image quality assessment metric in frequency domain , author=. SPIC , volume=. 2021 , publisher=

    2021

  129. [137]

    TCSVT , volume=

    Underwater image enhancement quality evaluation: Benchmark dataset and objective metric , author=. TCSVT , volume=. 2022 , publisher=

    2022

  130. [138]

    IJCV , volume=

    Learning to prompt for vision-language models , author=. IJCV , volume=. 2022 , publisher=

    2022

  131. [139]

    CVPR , pages=

    Conditional prompt learning for vision-language models , author=. CVPR , pages=

  132. [140]

    NeurIPS , pages=

    Bootstrap your own latent-a new approach to self-supervised learning , author=. NeurIPS , pages=

  133. [141]

    ICML , pages=

    Curriculum learning , author=. ICML , pages=

  134. [142]

    Remote Sens

    Underwater image restoration via contrastive learning and a real-world dataset , author=. Remote Sens. , volume=. 2022 , publisher=

    2022

  135. [143]

    CVPR , pages=

    Denseclip: Language-guided dense prediction with context-aware prompting , author=. CVPR , pages=

  136. [144]

    TIM , volume=

    Natural language processing: State of the art, current trends and challenges , author=. TIM , volume=. 2023 , publisher=

    2023

  137. [145]

    TGRS , volume=

    Human Perceptual Quality Driven Underwater Image Enhancement Framework , author=. TGRS , volume=. 2022 , publisher=

    2022

  138. [146]

    arXiv preprint arXiv:1906.06819 , year=

    A fusion adversarial underwater image enhancement network with a public test dataset , author=. arXiv preprint arXiv:1906.06819 , year=

    1906 arXiv

  139. [147]

    CVPR , pages=

    The unreasonable effectiveness of deep features as a perceptual metric , author=. CVPR , pages=

  140. [148]

    TIP , volume=

    An underwater color image quality evaluation metric , author=. TIP , volume=. 2015 , publisher=

    2015

  141. [149]

    JOE , volume=

    Human-visual-system-inspired underwater image quality measures , author=. JOE , volume=. 2015 , publisher=

    2015

  142. [150]

    ICCVW , pages=

    Transmission estimation in underwater single images , author=. ICCVW , pages=

  143. [151]

    TIP , volume=

    Underwater image enhancement via minimal color loss and locally adaptive contrast enhancement , author=. TIP , volume=. 2022 , publisher=

    2022

  144. [152]

    TCSVT , volume=

    Subjective and objective de-raining quality assessment towards authentic rain image , author=. TCSVT , volume=. 2020 , publisher=

    2020

  145. [153]

    PUGAN: Physical Model-Guided Underwater Image Enhancement Using GAN With Dual-Discriminators , year=

    Cong, Runmin and Yang, Wenyu and Zhang, Wei and Li, Chongyi and Guo, Chun-Le and Huang, Qingming and Kwong, Sam , journal=. PUGAN: Physical Model-Guided Underwater Image Enhancement Using GAN With Dual-Discriminators , year=

  146. [154]

    Islam, Md Jahidul and Wang, Ruobing and Sattar, Junaed , journal=

  147. [155]

    TGRS , volume=

    Ren, Tingdi and Xu, Haiyong and Jiang, Gangyi and Yu, Mei and Zhang, Xuan and Wang, Biao and Luo, Ting , title=. TGRS , volume=. 2022 , doi=

    2022

  148. [156]

    2023 , volume=

    Yan, Haorui and Zhang, Zhenwei and Xu, Jing and Wang, Tingting and An, Ping and Wang, Aobo and Duan, Yuping , journal=. 2023 , volume=

    2023

  149. [157]

    Iterative Prompt Learning for Unsupervised Backlit Image Enhancement , year=

    Liang, Zhexin and Li, Chongyi and Zhou, Shangchen and Feng, Ruicheng and Loy, Chen Change , booktitle=. Iterative Prompt Learning for Unsupervised Backlit Image Enhancement , year=

  150. [158]

    Unknown , year=

    Image enhancement by histogram transformation , author=. Unknown , year=

  151. [159]

    Underwater Image Enhancement Quality Evaluation: Benchmark Dataset and Objective Metric , year=

    Jiang, Qiuping and Gu, Yuese and Li, Chongyi and Cong, Runmin and Shao, Feng , journal=. Underwater Image Enhancement Quality Evaluation: Benchmark Dataset and Objective Metric , year=

  152. [160]

    Lin, Yupei and Xian, Xiaoyu and Shi, Yukai and Lin, Liang , journal=

  153. [161]

    ACM MM , pages=

    Exploring negatives in contrastive learning for unpaired image-to-image translation , author=. ACM MM , pages=

  154. [162]

    ECCV , pages=

    Perceptual losses for real-time style transfer and super-resolution , author=. ECCV , pages=. 2016 , organization=

    2016

  155. [163]

    2013 , publisher=

    Paull, Liam and Saeedi, Sajad and Seto, Mae and Li, Howard , journal=. 2013 , publisher=

    2013

  156. [164]

    2021 , publisher=

    Cong, Runmin and Zhang, Yumo and Fang, Leyuan and Li, Jun and Zhao, Yao and Kwong, Sam , journal=. 2021 , publisher=

    2021

  157. [165]

    JOE , volume=

    A reinforcement learning paradigm of configuring visual enhancement for object detection in underwater scenes , author=. JOE , volume=. 2023 , publisher=

    2023

  158. [166]

    SPIC , volume=

    Diving deeper into underwater image enhancement: A survey , author=. SPIC , volume=. 2020 , publisher=

    2020

  159. [167]

    Underwater Image Enhancement via Adaptive Group Attention-Based Multiscale Cascade Transformer , year=

    Huang, Zhixiong and Li, Jinjiang and Hua, Zhen and Fan, Linwei , journal=. Underwater Image Enhancement via Adaptive Group Attention-Based Multiscale Cascade Transformer , year=

  160. [168]

    Multiscale Underwater Image Enhancement in RGB and HSV Color Spaces , year=

    Liu, Chufan and Shu, Xin and Pan, Lei and Shi, Jinlong and Han, Bin , journal=. Multiscale Underwater Image Enhancement in RGB and HSV Color Spaces , year=

  161. [169]

    ECCV , pages=

    Finding Visual Task Vectors , author=. ECCV , pages=

  162. [170]

    NeurIPS , pages=

    Visual prompting via image inpainting , author=. NeurIPS , pages=

  163. [171]

    CVPR , pages=

    Low-light image enhancement via structure modeling and guidance , author=. CVPR , pages=

  164. [172]

    CVPR , pages=

    Efficient and explicit modelling of image hierarchies for image restoration , author=. CVPR , pages=

  165. [173]

    CVPR , pages=

    Imagen editor and editbench: Advancing and evaluating text-guided image inpainting , author=. CVPR , pages=

  166. [174]

    IJCV , pages=

    Exploiting diffusion prior for real-world image super-resolution , author=. IJCV , pages=

  167. [175]

    CVPR , year =

    Li, Bingchen and Li, Xin and Zhu, Hanxin and Jin, Yeying and Feng, Ruoyu and Zhang, Zhizheng and Chen, Zhibo , title =. CVPR , year =

  168. [176]

    CVPR , year =

    Wu, Rongyuan and Yang, Tao and Sun, Lingchen and Zhang, Zhengqiang and Li, Shuai and Zhang, Lei , title =. CVPR , year =

  169. [177]

    ICCV , year =

    Chen, Wei-Ting and Fang, Hao-Yu and Hsieh, Cheng-Lin and Tsai, Cheng-Che and Chen, I-Hsiang and Ding, Jian-Jiun and Kuo, Sy-Yen , title =. ICCV , year =

  170. [178]

    JOE , volume=

    Human-visual-system-inspired underwater image quality measures , author=. JOE , volume=

  171. [179]

    TIP , volume=

    An underwater color image quality evaluation metric , author=. TIP , volume=

  172. [180]

    and Shechtman, Eli and Wang, Oliver , title =

    Zhang, Richard and Isola, Phillip and Efros, Alexei A. and Shechtman, Eli and Wang, Oliver , title =. CVPR , pages=

  173. [181]

    NeurIPS , pages=

    Promptrestorer: A prompting image restoration method with degradation perception , author=. NeurIPS , pages=

  174. [182]

    IEEE Access , volume=

    Always Clear Days: Degradation Type and Severity Aware All-in-One Adverse Weather Removal , author=. IEEE Access , volume=

  175. [183]

    An Underwater Image Enhancement Benchmark Dataset and Beyond , year=

    Li, Chongyi and Guo, Chunle and Ren, Wenqi and Cong, Runmin and Hou, Junhui and Kwong, Sam and Tao, Dacheng , journal=. An Underwater Image Enhancement Benchmark Dataset and Beyond , year=

  176. [184]

    IJCV , pages=

    Gridformer: Residual dense transformer with grid structure for image restoration in adverse weather conditions , author=. IJCV , pages=

  177. [185]

    ECCV , pages=

    InstructIR: High-Quality Image Restoration Following Human Instructions , author=. ECCV , pages=

  178. [186]

    ECCV , pages=

    MambaIR: A Simple Baseline for Image Restoration with State-Space Model , author=. ECCV , pages=

  179. [187]

    AAAI , pages=

    Adaptive dynamic filtering network for image denoising , author=. AAAI , pages=

  180. [188]

    ICCV , pages=

    Diffir: Efficient diffusion model for image restoration , author=. ICCV , pages=

  181. [189]

    CVPR , pages=

    Uretinex-net: Retinex-based deep unfolding network for low-light image enhancement , author=. CVPR , pages=

  182. [190]

    ECCV , pages=

    Learning enriched features for real image restoration and enhancement , author=. ECCV , pages=

  183. [191]

    AFD-Former: A Hybrid Transformer With Asymmetric Flow Division for Synthesized View Quality Enhancement , year=

    Zhang, Xu and Cai, Nian and Zhang, Huan and Zhang, Yun and Di, Jianglei and Lin, Weisi , journal=. AFD-Former: A Hybrid Transformer With Asymmetric Flow Division for Synthesized View Quality Enhancement , year=

  184. [192]

    CVPR , pages=

    Deblurgan: Blind motion deblurring using conditional adversarial networks , author=. CVPR , pages=

  185. [193]

    ECCV , pages=

    Stripformer: Strip transformer for fast image deblurring , author=. ECCV , pages=

  186. [194]

    ICCV , pages=

    Retinexformer: One-Stage Retinex-Based Transformer for Low-Light Image Enhancement , author=. ICCV , pages=

  187. [195]

    NeurIPS , pages=

    Hierarchical Integration Diffusion Model for Realistic Image Deblurring , author=. NeurIPS , pages=

  188. [196]

    TIP , year=

    Vision Transformers for Single Image Dehazing , author=. TIP , year=

  189. [197]

    CVPR , month=

    Chen, Xiang and Li, Hao and Li, Mingqiang and Pan, Jinshan , title=. CVPR , month=. 2023 , pages=

    2023

  190. [198]

    TPAMI , volume=

    Image Restoration via Frequency Selection , author=. TPAMI , volume=

  191. [199]

    Kodak lossless true color image suite , author =

  192. [200]

    CVPR , pages=

    Reproducible scaling laws for contrastive language-image learning , author=. CVPR , pages=

  193. [201]

    NeurIPS , pages=

    DataComp: In Search of the Next Generation of Multimodal Datasets , author=. NeurIPS , pages=

  194. [202]

    ICLR , year=

    Very Deep Convolutional Networks for Large-Scale Image Recognition , author=. ICLR , year=

  195. [203]

    ECCV , pages=

    Visual prompt tuning , author=. ECCV , pages=

  196. [204]

    CVPR , pages=

    Diversity-Aware Meta Visual Prompting , author=. CVPR , pages=

  197. [205]

    CVPR , pages=

    Explicit visual prompting for low-level structure segmentations , author=. CVPR , pages=

  198. [206]

    ACL , pages=

    P-Tuning: Prompt Tuning Can Be Comparable to Fine-Tuning Across Scales and Tasks , author=. ACL , pages=

  199. [207]

    NeurIPS , pages=

    Language models are few-shot learners , author=. NeurIPS , pages=

  200. [208]

    CVPR , pages=

    Prompt distribution learning , author=. CVPR , pages=

  201. [209]

    arXiv preprint arXiv:2401.06013 , year=

    Surgical-DINO: Adapter Learning of Foundation Model for Depth Estimation in Endoscopic Surgery , author=. arXiv preprint arXiv:2401.06013 , year=

    arXiv

  202. [210]

    NeurIPS , pages=

    A Tale of Two Features: Stable Diffusion Complements DINO for Zero-Shot Semantic Correspondence , author=. NeurIPS , pages=

  203. [211]

    CVPR , pages=

    Cris: Clip-driven referring image segmentation , author=. CVPR , pages=

  204. [212]

    CVPR , pages=

    Hairclip: Design your hair by text and reference image , author=. CVPR , pages=

  205. [213]

    CVPR , pages=

    All-in-one image restoration for unknown degradations using adaptive discriminative filters for specific degradations , author=. CVPR , pages=

  206. [214]

    CVPR , pages=

    All in one bad weather removal using architectural search , author=. CVPR , pages=

  207. [215]

    CVPR , pages=

    Learning multiple adverse weather removal via two-stage knowledge learning and multi-contrastive regularization: Toward a unified model , author=. CVPR , pages=

  208. [216]

    CVPR , pages=

    Learning Weather-General and Weather-Specific Features for Image Restoration Under Multiple Adverse Weather Conditions , author=. CVPR , pages=

  209. [217]

    ICLR , year=

    Controlling Vision-Language Models for Universal Image Restoration , author=. ICLR , year=

  210. [218]

    Exploring the Application of Large-Scale Pre-Trained Models on Adverse Weather Removal , year=

    Tan, Zhentao and Wu, Yue and Liu, Qiankun and Chu, Qi and Lu, Le and Ye, Jieping and Yu, Nenghai , journal=. Exploring the Application of Large-Scale Pre-Trained Models on Adverse Weather Removal , year=

  211. [219]

    arXiv preprint arXiv:2306.13653 , year=

    ProRes: Exploring Degradation-aware Visual Prompt for Universal Image Restoration , author=. arXiv preprint arXiv:2306.13653 , year=

    arXiv

  212. [220]

    ICCV , pages=

    Unsupervised Image Denoising in Real-World Scenarios via Self-Collaboration Parallel Generative Adversarial Branches , author=. ICCV , pages=

  213. [221]

    CVPR , pages=

    Unpaired deep image deraining using dual contrastive learning , author=. CVPR , pages=

  214. [222]

    CVPR , pages=

    Density-aware single image de-raining using a multi-stream dense network , author=. CVPR , pages=

  215. [223]

    CVPR , pages=

    Efficient Frequency Domain-based Transformers for High-Quality Image Deblurring , author=. CVPR , pages=

  216. [224]

    CVPR , pages=

    Multi-scale progressive fusion network for single image deraining , author=. CVPR , pages=

  217. [225]

    CVPR , pages=

    Semi-supervised transfer learning for image rain removal , author=. CVPR , pages=

  218. [226]

    CVPR , pages=

    Uncertainty guided multi-scale residual learning-using a cycle spinning cnn for single image de-raining , author=. CVPR , pages=

  219. [227]

    ICCV , pages=

    Aod-net: All-in-one dehazing network , author=. ICCV , pages=

  220. [228]

    TIP , volume=

    Dehazenet: An end-to-end system for single image haze removal , author=. TIP , volume=

  221. [229]

    CVPR , pages=

    Single image super-resolution from transformed self-exemplars , author=. CVPR , pages=

  222. [230]

    TIP , volume=

    Beyond a gaussian denoiser: Residual learning of deep cnn for image denoising , author=. TIP , volume=

  223. [231]

    TIP , volume=

    FFDNet: Toward a fast and flexible solution for CNN-based image denoising , author=. TIP , volume=

  224. [232]

    arXiv preprint arXiv:2312.01677 , year=

    Multi-task Image Restoration Guided By Robust DINO Features , author=. arXiv preprint arXiv:2312.01677 , year=

    arXiv

  225. [233]

    ICLR , pages=

    Learning transferable visual models from natural language supervision , author=. ICLR , pages=

  226. [234]

    ICCV , pages=

    Emerging properties in self-supervised vision transformers , author=. ICCV , pages=

  227. [235]

    TMLR , year=

    DINOv2: Learning Robust Visual Features without Supervision , author=. TMLR , year=

  228. [236]

    ICCV , pages=

    Iterative prompt learning for unsupervised backlit image enhancement , author=. ICCV , pages=

  229. [237]

    TIP , volume=

    Benchmarking single-image dehazing and beyond , author=. TIP , volume=

  230. [238]

    CVPR , pages=

    Hinet: Half instance normalization network for image restoration , author=. CVPR , pages=

  231. [239]

    ECCV , pages=

    Simple baselines for image restoration , author=. ECCV , pages=

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.