Pith. sign in

REVIEW 6 major objections 7 minor 81 references

Degradation-Aware Prompt Learning with Cross-Modal Compensation for Adverse Weather Removal

T0 review · 6 major / 7 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 This paper proposes DCMPC-Net, an all-in-one adverse-weather removal network whose degradation-aware cross-modal prompts—built by cross-attending frozen language-model text embeddings with image features—outperform prior task-specific and…

desk verdict Solid all-in-one weather removal with a plausible cross-modal prompt design, but the semantic story is under-supported and the added FFT loss muddies the headline margin. read the letter →

arxiv 2608.06939 v2 pith:JWJP6N57 submitted 2026-08-07 cs.CV

Wanshu Fan , Yunzhe Zhang , Yue Shen , Liyan Wang , Jing Qin , Kin-Man Lam , Cong Wang , Jinshan Pan This is my paper · ORCID
classification cs.CV
keywords adverseweatherremovalall-in-oneimagerestorationcross-modalpromptlearningvision-languagemodeldegradation-awarepromptsattentionalignmentsnowraindrop
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 proposes DCMPC-Net, a single network for removing rain, snow, fog, and raindrops that conditions a shared restoration backbone with degradation-aware prompts built from both text and image cues. Its central claim is that coupling language-derived semantic descriptions of weather with visual features, then injecting the resulting prompts into the decoder through attention alignment, systematically improves restoration accuracy over both the backbone alone and prior prompt-based all-in-one methods. The reported result is an average 34.03 dB PSNR and 0.9486 SSIM across Snow100K-S, Snow100K-L, Outdoor-Rain, and RainDrop, 0.27 dB above the second-best compared method. If true, the paper shows that semantic knowledge from a frozen vision-language model can be made spatially useful for low-level restoration without changing the computation path or routing experts.

What carries the argument

The central object is the Cross-Modal Prompt Generator (CMPG), which produces a degradation-aware cross-modal prompt $P_{da}$ by cross-attention between a frozen text encoder's embedding $T_{emb}$ (query) and visual features $I_{emb}$ (key and value): $P_{da} = V \cdot \mathrm{Softmax}(K \cdot Q / \beta)$. This latent prompt is injected into the decoder through the Prompt-Guided Attention Alignment Module (PGAAM), whose dual cross-attention establishes spatial correspondence between semantic cues and degraded regions; the Dual Feature Compensation Module (DFCM) additionally uses the channel-wise max-minus-min of the input as a compensation map to reinforce structures. The prompt is the load-bearing mechanism that carries semantic degradation knowledge into feature refinement.

What would settle it

Run the full model with the text descriptions replaced by deliberately generic or wrong captions (for example 'clear sunny scene' for every rainy input) while keeping all modules identical; if the average PSNR does not fall toward the no-prompt baseline, then the cross-modal semantic content is not what drives the gain. Alternatively, compute prompts live at test time instead of loading cached embeddings and check whether accuracy drops, which would reveal whether the cached-prompt protocol is masking the online cost of semantic extraction.

Watch

Extended reading notes

Core claim

DCMPC-Net discovers that degradation-aware cross-modal prompts—latent representations formed by cross-attending text embeddings from a frozen language model with degraded image features—can serve as effective feature-level conditioning signals in a unified adverse-weather removal network. These prompts are aligned to degraded regions by a Prompt-Guided Attention Alignment Module, while a Dual Feature Compensation Module separates degradation artifacts from scene structure to preserve fine detail. The paper argues this design outperforms image-level visual prompts, text-only cues, and routing-based language priors because it retains spatial correspondence and provides continuous modulation rather than discrete expert selection. The strongest quantitative claim is the best average PSNR and SSIM across the four benchmarks, with the largest gains on Outdoor-Rain where degradation is spatially uneven.

Load-bearing premise

The whole gain rests on the frozen vision-language model, prompted with a fixed question, producing text descriptions that genuinely encode the local weather degradation, and on those semantics surviving when the prompts are cached offline and simply loaded during training and testing.

Editorial extensions

If this is right

  • One trained model can handle snow, rain&fog, and raindrops together, outperforming task-specific and previous all-in-one methods on the four evaluated benchmarks.
  • Language-derived semantics can be injected as continuous feature-conditioning signals without expert routing or explicit degradation maps, keeping the restoration computation path fixed.
  • Each component contributes additively: removing the cross-modal prompt generator costs about 0.2 dB on Outdoor-Rain, showing that prompt semantics matter at the feature level rather than as a mere auxiliary description.
  • Because prompts are precomputed and cached, the expensive frozen semantic-prior stage can run offline, leaving the restoration-stage runtime at 0.08 seconds per 256x256 image.
  • The ablation against LDR indicates that converting vision-language priors into a continuous prompt is competitive with converting them into routing scores, with DCMPC-Net reporting higher average SSIM across the three shared test sets.

Reading between the lines

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

  • Our inference: if the frozen model's descriptions become generic or misaligned with local degradation, the cross-modal prompt should degrade toward a static token; this is testable by ablating the prompt query across levels of vagueness and predicting a monotonic drop in the Outdoor-Rain gain.
  • Our inference: the same conditioning mechanism could transfer to other spatially heterogeneous degradations such as reflections, shadows, or local blur, where text describes the artifact type but attention must localize it; the paper does not test these settings.
  • Our inference: the cached-prompt protocol means real-time cost is decoupled from the frozen model, so a natural next step—not explored here—is learning a lightweight prompt estimator that replaces the offline extraction pass.
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

6 major / 7 minor

Summary. The paper proposes DCMPC-Net, an all-in-one adverse-weather restoration network built on the Histoformer backbone, augmented by a Cross-Modal Prompt Generator (CMPG) that uses a frozen vision-language model and text encoder to produce degradation-aware prompts, plus a Prompt-Guided Attention Alignment Module (PGAAM) and a Dual Feature Compensation Module (DFCM). The method is evaluated on Snow100K-S/L, OutdoorRain, and RainDrop, reporting the best average PSNR/SSIM (34.03 dB / 0.9486) among the compared methods, with ablations over the proposed components, semantic sources, and module placements.

Significance. If the central attribution is correct, the paper makes a useful empirical contribution by showing that frozen VLM-derived semantic prompts can be injected as feature-level conditioning in a unified restoration backbone, and the code release and explicit failure-case discussion are positive features. The average gain of 0.27 dB over the second-best method is modest but non-negligible. However, the evidence for the degradation-aware semantic mechanism is not yet isolated: an extra FFT loss is used without ablation, a constant-text control is missing, the unimodal text branch is sometimes worse than the backbone, and the formal equations for the core cross-attention are dimensionally invalid. These issues must be addressed before the specific semantic claim can be accepted.

major comments (6)
  1. [Section III-B, Eqs. (4)-(5)] Equations (4)-(5) are not algebraically consistent with the stated tensor shapes. With T_emb ∈ R^{L×C} and I_emb ∈ R^{H×W×C}, the projections cannot all produce Q, K, V ∈ R^{H×W×C}; moreover, P_da = V·softmax(K·Q/β) is not a valid cross-attention contraction because K·Q does not have compatible dimensions (no transposition is indicated, and the text sequence length L is not spatial H×W). Since CMPG is the central novelty, the paper is not self-contained without corrected equations or an explicit pointer to the exact implementation.
  2. [Section IV-C and IV-E, Eq. (13)] Section IV-C states that the FFT loss is incorporated 'in addition to those used in the baseline setting,' but no experiment removes L_fre. Consequently, the 0.27 dB average margin over CyclicPrompt in Table I, and the improvements of the full model over Histoformer in Table V, cannot be attributed solely to CMPG/PGAAM/DFCM. Please add an ablation with and without L_fre under otherwise identical settings, and ensure that compared baselines are trained with the same loss.
  3. [Section IV-E.2, Tables V-VI] The current ablations do not isolate the weather-specific semantic content of the LLaMA descriptions. In Table V, replacing CMPG with text features alone yields 31.95 dB on OutdoorRain and 32.96 dB on RainDrop, below the Histoformer backbone (32.08 and 33.06), so unimodal text guidance is not beneficial without the rest of the pipeline. In Table VI, learnable queries are worse than the frozen-text branch, but this only shows that a frozen text initialization helps. A control experiment in which CMPG receives a fixed, semantically vacuous text description (e.g., 'a photo') with all other components and losses unchanged is needed to determine whether the degradation-specific wording matters. Without this control, the 'degradation-aware semantic' explanation is not load-bearing.
  4. [Section IV-D.1, Tables I-II] The paper claims to 'outperform state-of-the-art methods' in the abstract and Section I, but Table I shows DCMPC-Net does not achieve the best per-dataset PSNR on OutdoorRain (32.49 vs 32.81 for CyclicPrompt) or on Snow100K-L (32.35 vs 32.37 for T3-DiffWeather). Table II similarly shows LDR has higher PSNR on Snow100K-L (32.47 vs 32.35) and RainDrop (33.83 vs 33.08). The supported claim is best average PSNR/SSIM, not consistent superiority; please rephrase the claims to match the evidence.
  5. [Section IV-E.2, Table VII] Table VII reports Histoformer Snow100K-S SSIM as 0.9689, while Tables I and V report 0.9658 for the same configuration. This changes the apparent margin of the full model (0.9692 vs 0.9658 = +0.0034 in Table I, but +0.0003 against 0.9689 in Table VII) and complicates the ablation interpretation. The table also groups 'BLIP & CLIP' into one row despite the text saying results are derived from BLIP, CLIP, or no-prompt configurations; please report these variants separately and correct the inconsistent number.
  6. [Section III-B.1, Eq. (1)] The paper describes LLaMA [35] as a 'pre-trained vision-language model' and writes T_text = PVL(Q_t, I_LQ), feeding the input image to LLaMA. LLaMA is a language model and cannot process image inputs unless a multimodal adapter is used (e.g., LLaVA). Please specify the exact VLM and image-input interface, and correct the related statement that it was pre-trained on 15 trillion multimodal tokens.
minor comments (7)
  1. [Table I] The header 'Histormer' is a typo for 'Histoformer'.
  2. [Section IV-C] Please state which losses constitute the baseline setting so that the FFT-loss addition is precisely defined.
  3. [Section III-E, Eq. (16)] Please define F(·) as the Fourier transform operator and specify the normalization used in the frequency-domain loss.
  4. [Section III-D, Eq. (12) and Fig. 4] The variables y_t and y_{t+1} appear in the DFCM equations and in Fig. 4 but are not formally introduced in the text; please define them.
  5. [Table VI] Many configurations share identical Snow100K-S PSNR/SSIM values (e.g., 38.00/0.9683), which may reflect rounding at 0.01 dB; consider reporting more significant digits to make the ablations informative.
  6. [Table VII] The row label 'BLIP & CLIP' is ambiguous; please report BLIP and CLIP separately, along with the prompt template used for each.
  7. [Section IV-D.1] For comparisons with LDR computed from the official repository, please confirm that the same evaluation protocol (Y-channel PSNR/SSIM, border cropping, and full-resolution restoration) was used as for the other methods.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: central claim is an external-benchmark comparison; self-citations are contextual, not load-bearing.

full rationale

The paper's central claim is that DCMPC-Net outperforms state-of-the-art methods, and this is established by comparing trained models against external benchmark test sets (Snow100K-S/L, Outdoor-Rain, RainDrop) with fixed ground truths, plus component ablations (Tables IV-X) that remove or replace CMPG, PGAAM, DFCM, and semantic sources. None of these reported quantities is constructed from the final PSNR/SSIM statistics. The prompt embeddings are precomputed from degraded inputs with a fixed textual query under the cached-prompt protocol (Section IV-C) and are not fitted to the test metrics. The use of Histoformer as the backbone and the citation of PromptRestorer [20] for the prompt-driven attention paradigm involve co-authors of prior work, but these citations supply architectural lineage rather than the evidential load, and all comparison numbers are measured on independent benchmarks. The added FFT loss (Section IV-C) is a possible confound for attributing the 0.27 dB average margin to the proposed modules alone, but it is a training objective, not a fitted parameter that makes the output equal to the target by construction. The dimension inconsistency in Eqs. (4)-(5) is a formal specification error, not a circular reduction. No passage asserts a circular step or an omitted proof that would change this verdict.

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

The central claim is empirical; no physical constants or fitted parameters appear. The load-bearing design inputs are the fixed prompt text, the loss weighting, and the assumption that frozen semantic priors remain useful when cached.

free parameters (2)
  • Loss term weights = not reported; Eq. (13) sums L1, Lcor, Lfre without explicit weights
    The combined loss in Eq. (13) sets all three terms to equal weight. The relative scales of pixel L1, correlation, and FFT losses are not discussed, so the effective weighting is an unstated design choice that affects optimization.
  • Fixed textual query = Please describe the type of weather, intensity, and obscured areas
    This hand-written prompt is the only semantic input to the frozen PVL. Its wording is not varied or ablated, yet the cross-modal prompts and all downstream results depend on it.
assumptions (3)
  • domain assumption Frozen LLaMA and Multilingual-E5 embeddings encode degradation semantics useful for restoration.
    The entire CMPG rests on this. If the generated text descriptions are uninformative or misaligned with visual degradations, the cross-modal prompts add no useful signal. Invoked in Section III-B.
  • domain assumption The cached-prompt protocol is equivalent to online PVL inference during training and testing.
    Prompts are precomputed offline and loaded during training/testing (Section IV-C). This assumes no distribution shift and that the frozen models need not be fine-tuned.
  • domain assumption Evaluation protocol of compared methods is consistent.
    Numbers for competitors are taken 'as reported in the original papers' (Section IV-D.1), and LDR is evaluated only from official repo outputs. Differences in training data, crops, and metric computation can affect comparisons.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Degradation-Aware Prompt Learning with Cross-Modal Compensation for Adverse Weather Removal." pith.science (2026). https://pith.science/paper/JWJP6N57

@misc{pith2026260806939,
  author       = {Pith},
  title        = {Pith review of: Degradation-Aware Prompt Learning with Cross-Modal Compensation for Adverse Weather Removal},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JWJP6N57}},
  note         = {Machine review of arXiv:2608.06939}
}
read the original abstract

Adverse weather causes diverse and complex image degradations, severely compromising the reliability of computer vision systems. Existing all-in-one restoration models attempt to address multiple degradation types within a unified framework, but often lack explicit spatial and semantic modeling of degradation characteristics, limiting their adaptability to diverse weather conditions. To address this limitation, we propose a Degradation-Aware Cross-Modal Prompt Compensation Network (DCMPC-Net) that leverages cross-modal degradation cues from a pretrained vision-language model to condition restoration features within a unified backbone. Specifically, our DCMPC-Net mainly consists of the Cross-Modal Prompt Generator (CMPG), Prompt-Guided Attention Alignment Module (PGAAM), and Dual Feature Compensation Module (DFCM). The CMPG integrates textual embeddings with visual features to produce degradation-aware prompts that encode degradation-related semantic and contextual cues. These prompts are injected into the decoder via a PGAAM, which adaptively aligns semantic information with degraded regions to facilitate context-aware restoration. To further enhance structural fidelity, DFCM is introduced that disentangles degradation artifacts from scene structures, thereby improving the reconstruction of fine textures and detailed content. By integrating cross-modal semantic guidance with spatial alignment and structural enhancement, DCMPC-Net achieves robust and perceptually consistent restoration across diverse weather conditions. Extensive experiments show that DCMPC-Net outperforms state-of-the-art methods in both task-specific and unified settings, achieving superior accuracy and visual fidelity.

Figures

Figures reproduced from arXiv: 2608.06939 by the authors.

Figure 1
Figure 1. Conceptual comparison of restoration frameworks for [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 5
Figure 5. Ours [PITH_FULL_IMAGE:figures/full_fig_p002_5.png] view at source ↗
Figure 2
Figure 2. A comparative analysis demonstrates that our proposed approach [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figures from the paper (21 more)
Figure 7
Figure 7. Figure 7: Overview of the proposed DCMPC-Net architecture. The framework consists of three core components: Cross-Modal [PITH_FULL_IMAGE:figures/full_fig_p004_7.png]
Figure 3
Figure 3. Figure 3: Overview of the proposed DCMPC-Net architecture. The framework consists of three core components: Cross-Modal Prompt Generator (CMPG), [PITH_FULL_IMAGE:figures/full_fig_p004_3.png]
Figure 8
Figure 8. Figure 8: Visualization of Pda, Fmap, yt, and yt+1. From Pda to Fmap, the degraded regions become more prominent, demonstrating the effectiveness of the proposed PGAAM in highlighting degradation cues. Furthermore, the transition from yt to yt+1 reveals enhanced structural detai…
Figure 4
Figure 4. Figure 4: Visualization of Pda, Fmap, yt, and yt+1. From Pda to Fmap, the degraded regions become more prominent, demonstrating the effectiveness of the proposed PGAAM in highlighting degradation cues. Furthermore, the transition from yt to yt+1 reveals enhanced structural detai…
Figure 9
Figure 9. Figure 9: Visual comparison with SOTA methods on the Snow-100k test set [ [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 5
Figure 5. Figure 5: Visual comparison with SOTA methods on the Snow-100k test set [ [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 10
Figure 10. Figure 10: Visual comparison with SOTA methods on the OutdoorRain test set [ [PITH_FULL_IMAGE:figures/full_fig_p009_10.png]
Figure 6
Figure 6. Figure 6: Visual comparison with SOTA methods on the OutdoorRain test set [ [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 11
Figure 11. Figure 11: Visual comparison with SOTA methods on the RainDrop test set [ [PITH_FULL_IMAGE:figures/full_fig_p009_11.png]
Figure 7
Figure 7. Figure 7: Visual comparison with SOTA methods on the RainDrop test set [ [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 12
Figure 12. Figure 12: Visual comparison with SOTA methods on the Real-Snow test set [ [PITH_FULL_IMAGE:figures/full_fig_p010_12.png]
Figure 8
Figure 8. Figure 8: Visual comparison with SOTA methods on the Real-Snow test set [ [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 13
Figure 13. Figure 13: Visual comparison of models with different configurations. Incorporating CMPG, PGAAM, and DFCM enables [PITH_FULL_IMAGE:figures/full_fig_p011_13.png]
Figure 9
Figure 9. Figure 9: Visual comparison of models with different configurations. Incorporating CMPG, PGAAM, and DFCM enables DCMPC-Net to produce results with [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 25
Figure 25. Figure 25: (a) Input [PITH_FULL_IMAGE:figures/full_fig_p013_25.png]
Figure 10
Figure 10. Figure 10: Visual comparison between different pre-trained vision language [PITH_FULL_IMAGE:figures/full_fig_p011_10.png]
Figure 18
Figure 18. Figure 18: (a) Input [PITH_FULL_IMAGE:figures/full_fig_p013_18.png]
Figure 12
Figure 12. Figure 12: Visual comparison between w/o RB and w/o CB on the Snow100K [PITH_FULL_IMAGE:figures/full_fig_p012_12.png]
Figure 35
Figure 35. Figure 35: Input [PITH_FULL_IMAGE:figures/full_fig_p014_35.png]
Figure 13
Figure 13. Figure 13: Example of object detection results before and after adverse weather [PITH_FULL_IMAGE:figures/full_fig_p013_13.png]
Figure 14
Figure 14. Figure 14: Failure case of DCMPC-Net under heavy rain conditions. When [PITH_FULL_IMAGE:figures/full_fig_p013_14.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

81 extracted references · 4 canonical work pages

  1. [35]

    The llama 3 herd of models,

    A. Dubey, A. Jauhri, and et al., “The llama 3 herd of models,” arXiv:2407.21783, 2024

    arXiv 2024

  2. [1]

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

    R. Qian, R. T. Tan, W. Yang, J. Su, and J. Liu, “Attentive generative adversarial network for raindrop removal from a single image,” inCVPR, 2018

    2018

  3. [2]

    Dcsfn: Deep cross-scale fusion network for single image rain removal,

    C. Wang, X. Xing, Y . Wu, Z. Su, and J. Chen, “Dcsfn: Deep cross-scale fusion network for single image rain removal,” inACM MM, 2020, pp. 1643–1651

    2020

  4. [3]

    Joint self-attention and scale- aggregation for self-calibrated deraining network,

    C. Wang, Y . Wu, Z. Su, and J. Chen, “Joint self-attention and scale- aggregation for self-calibrated deraining network,” inACM MM, 2020, pp. 2517–2525

    2020

  5. [4]

    Online-updated high-order collaborative networks for single image deraining,

    C. Wang, J. Pan, and X. Wu, “Online-updated high-order collaborative networks for single image deraining,” inAAAI, 2022, pp. 2406–2413

    2022

  6. [5]

    Learning A sparse transformer network for effective image deraining,

    X. Chen, H. Li, M. Li, and J. Pan, “Learning A sparse transformer network for effective image deraining,” inCVPR, 2023

    2023

  7. [6]

    Explore internal and external similarity for single image deraining with graph neural networks,

    C. Wang, W. Wang, C. Yu, and J. Mu, “Explore internal and external similarity for single image deraining with graph neural networks,” in IJCAI, 2024, pp. 1371–1379

    2024

  8. [7]

    Progressive local and non-local interactive networks with deeply discriminative training for image deraining,

    C. Wang, L. Wang, J. Mu, C. Yu, and W. Wang, “Progressive local and non-local interactive networks with deeply discriminative training for image deraining,” inACM MM, 2024, pp. 10 326–10 335

    2024

Show all 81 references
  1. [8]

    Intra and inter parser- prompted transformers for effective image restoration,

    C. Wang, J. Pan, L. Wang, and W. Wang, “Intra and inter parser- prompted transformers for effective image restoration,” inAAAI, 2025

    2025

  2. [9]

    Gated fusion network for single image dehazing,

    W. Ren, L. Ma, J. Zhang, J. Pan, X. Cao, W. Liu, and M. Yang, “Gated fusion network for single image dehazing,” inCVPR, 2018

    2018

  3. [10]

    Selfpromer: Self-prompt dehazing transformers with depth-consistency,

    C. Wang, J. Pan, W. Lin, J. Dong, W. Wang, and X. Wu, “Selfpromer: Self-prompt dehazing transformers with depth-consistency,” inAAAI, 2024

    2024

  4. [11]

    Correlation matching transformation transformers for UHD image restoration,

    C. Wang, J. Pan, W. Wang, G. Fu, S. Liang, M. Wang, X. Wu, and J. Liu, “Correlation matching transformation transformers for UHD image restoration,” inAAAI, 2024, pp. 5336–5344. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 14

    2024

  5. [12]

    Neural discrimination-prompted transformers for efficient uhd image restoration and enhancement,

    C. Wang, J. Pan, L. Wang, W. Wang, and Y . Yang, “Neural discrimination-prompted transformers for efficient uhd image restoration and enhancement,”IJCV, vol. 134, no. 3, p. 84, 2026

    2026

  6. [13]

    Online rain/snow removal from surveillance videos,

    M. Li, X. Cao, Q. Zhao, L. Zhang, and D. Meng, “Online rain/snow removal from surveillance videos,”TIP, 2021

    2021

  7. [14]

    Deep dense multi-scale network for snow removal using semantic and depth priors,

    K. Zhang, R. Li, Y . Yu, W. Luo, and C. Li, “Deep dense multi-scale network for snow removal using semantic and depth priors,”TIP, 2021

    2021

  8. [15]

    All-in-one image restoration for unknown corruption,

    B. Li, X. Liu, P. Hu, Z. Wu, J. Lv, and X. Peng, “All-in-one image restoration for unknown corruption,” inCVPR, 2022

    2022

  9. [16]

    All in one bad weather removal using architectural search,

    R. Li, R. T. Tan, and L. Cheong, “All in one bad weather removal using architectural search,” inCVPR, 2020

    2020

  10. [17]

    Restormer: Efficient transformer for high-resolution image restoration,

    S. W. Zamir, A. Arora, S. Khan, M. Hayat, F. S. Khan, and M. Yang, “Restormer: Efficient transformer for high-resolution image restoration,” inCVPR, 2022

    2022

  11. [18]

    Learning a low-level vision generalist via visual task prompt,

    X. Chen, Y . Liu, Y . Pu, W. Zhang, J. Zhou, Y . Qiao, and C. Dong, “Learning a low-level vision generalist via visual task prompt,” inACM MM, 2024

    2024

  12. [19]

    All-in-one weather-degraded image restoration via adaptive degradation-aware self- prompting model,

    Y . Wen, T. Gao, Z. Li, J. Zhang, K. Zhang, and T. Chen, “All-in-one weather-degraded image restoration via adaptive degradation-aware self- prompting model,”TOM, 2024

    2024

  13. [20]

    Promptrestorer: A prompting image restoration method with degrada- tion perception,

    C. Wang, J. Pan, W. Wang, J. Dong, M. Wang, Y . Ju, and J. Chen, “Promptrestorer: A prompting image restoration method with degrada- tion perception,” inNIPS, 2023

    2023

  14. [21]

    Promptir: Prompting for all-in-one image restoration,

    V . Potlapalli, S. W. Zamir, S. H. Khan, and F. S. Khan, “Promptir: Prompting for all-in-one image restoration,” inNIPS, 2023

    2023

  15. [22]

    Adverse weather removal with codebook priors,

    T. Ye, S. Chen, J. Bai, J. Shi, C. Xue, J. Jiang, J. Yin, E. Chen, and Y . Liu, “Adverse weather removal with codebook priors,” inCVPR, 2023

    2023

  16. [23]

    Instructir: High-quality image restoration following human instructions,

    M. V . Conde, G. Geigle, and R. Timofte, “Instructir: High-quality image restoration following human instructions,” inECCV, 2024

    2024

  17. [24]

    Controlling vision-language models for multi-task image restoration,

    Z. Luo, F. K. Gustafsson, Z. Zhao, J. Sj ¨olund, and T. B. Sch ¨on, “Controlling vision-language models for multi-task image restoration,” inICLR, 2024

    2024

  18. [25]

    Dreamclear: High-capacity real-world image restoration with privacy- safe dataset curation,

    Y . Ai, X. Zhou, H. Huang, X. Han, Z. Chen, Q. You, and H. Yang, “Dreamclear: High-capacity real-world image restoration with privacy- safe dataset curation,” inNIPS, 2024

    2024

  19. [26]

    Multimodal prompt perceiver: Empower adaptiveness, generalizability and fidelity for all-in- one image restoration,

    Y . Ai, H. Huang, X. Zhou, J. Wang, and R. He, “Multimodal prompt perceiver: Empower adaptiveness, generalizability and fidelity for all-in- one image restoration,” inCVPR, 2024

    2024

  20. [27]

    Language-driven all-in-one adverse weather removal,

    H. Yang, L. Pan, Y . Yang, and W. Liang, “Language-driven all-in-one adverse weather removal,” inCVPR, 2024

    2024

  21. [28]

    Transweather: Transformer-based restoration of images degraded by adverse weather conditions,

    J. M. J. Valanarasu, R. Yasarla, and V . M. Patel, “Transweather: Transformer-based restoration of images degraded by adverse weather conditions,” inCVPR, 2022

    2022

  22. [29]

    Frequency-oriented efficient transformer for all-in-one weather-degraded image restoration,

    T. Gao, Y . Wen, K. Zhang, J. Zhang, T. Chen, L. Liu, and W. Luo, “Frequency-oriented efficient transformer for all-in-one weather-degraded image restoration,”TCSVT, 2023

    2023

  23. [30]

    Adair: Adaptive all-in-one image restoration via frequency mining and modulation,

    Y . Cui, S. W. Zamir, S. H. Khan, A. Knoll, M. Shah, and F. S. Khan, “Adair: Adaptive all-in-one image restoration via frequency mining and modulation,” inICLR, 2025

    2025

  24. [31]

    Restoring images in adverse weather conditions via histogram transformer,

    S. Sun, W. Ren, X. Gao, R. Wang, and X. Cao, “Restoring images in adverse weather conditions via histogram transformer,” inECCV, 2024

    2024

  25. [32]

    Moerl: When mixture-of-experts meet reinforcement learning for adverse weather image restoration,

    T. Wang, P. Xia, B. Li, P.-T. Jiang, Z. Kong, K. Zhang, T. Lu, and W. Luo, “Moerl: When mixture-of-experts meet reinforcement learning for adverse weather image restoration,” inICCV, 2025

    2025

  26. [33]

    Bert: Pre-training of deep bidirectional transformers for language understanding,

    J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “Bert: Pre-training of deep bidirectional transformers for language understanding,” inACL, 2019

    2019

  27. [34]

    Improving language understanding by generative pre-training,

    A. Radford, K. Narasimhan, T. Salimans, I. Sutskeveret al., “Improving language understanding by generative pre-training,” 2018

    2018

  28. [36]

    Learning transferable visual models from natural language supervi- sion,

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, G. Krueger, and I. Sutskever, “Learning transferable visual models from natural language supervi- sion,” inICML, 2021

    2021

  29. [37]

    BLIP-2: bootstrapping language-image pre-training with frozen image encoders and large language models,

    J. Li, D. Li, S. Savarese, and S. C. H. Hoi, “BLIP-2: bootstrapping language-image pre-training with frozen image encoders and large language models,” inICML, 2023

    2023

  30. [38]

    Deep visual-semantic alignments for generating image descriptions,

    A. Karpathy and L. Fei-Fei, “Deep visual-semantic alignments for generating image descriptions,” inCVPR, 2015

    2015

  31. [39]

    A global visual information intervention model for medical visual question answering,

    P. Peng, W. Fan, Y . Shen, X. Yang, and D. Zhou, “A global visual information intervention model for medical visual question answering,” Comput. Biol. Med., 2025

    2025

  32. [40]

    Improving image restoration through removing degradations in textual representations,

    J. Lin, Z. Zhang, Y . Wei, D. Ren, D. Jiang, Q. Tian, and W. Zuo, “Improving image restoration through removing degradations in textual representations,” inCVPR, 2024

    2024

  33. [41]

    Visual prompting via image inpainting,

    A. Bar, Y . Gandelsman, T. Darrell, A. Globerson, and A. Efros, “Visual prompting via image inpainting,”NIPS, 2022

    2022

  34. [42]

    Improving natural language processing tasks with human gaze-guided neural attention,

    E. Sood, S. Tannert, P. M ¨uller, and A. Bulling, “Improving natural language processing tasks with human gaze-guided neural attention,” NIPS, 2020

    2020

  35. [43]

    Exploiting cloze questions for few shot text classification and natural language inference,

    T. Schick and H. Sch ¨utze, “Exploiting cloze questions for few shot text classification and natural language inference,”arXiv:2001.07676, 2020

    2001 arXiv

  36. [44]

    Auto- prompt: Eliciting knowledge from language models with automatically generated prompts,

    T. Shin, Y . Razeghi, R. L. Logan IV , E. Wallace, and S. Singh, “Auto- prompt: Eliciting knowledge from language models with automatically generated prompts,”arXiv:2010.15980, 2020

    2010 arXiv

  37. [45]

    Visual prompt tuning,

    M. Jia, L. Tang, B.-C. Chen, C. Cardie, S. Belongie, B. Hariharan, and S.-N. Lim, “Visual prompt tuning,” inECCV, 2022

    2022

  38. [46]

    Prompt-in-prompt learning for universal image restoration,

    Z. Li, Y . Lei, C. Ma, J. Zhang, and H. Shan, “Prompt-in-prompt learning for universal image restoration,”arXiv:2312.05038, 2023

    2023 arXiv

  39. [47]

    Textual prompt guided image restoration,

    Q. Yan, A. Jiang, K. Chen, L. Peng, Q. Yi, and C. Zhang, “Textual prompt guided image restoration,”EAAI, 2025

    2025

  40. [48]

    Mrir: Integrating multimodal insights for diffusion-based realistic im- age restoration,

    Y . Zhang, H. Zhang, X. Chai, R. Xie, L. Song, and W. Zhang, “Mrir: Integrating multimodal insights for diffusion-based realistic im- age restoration,”TCSVT, 2025

    2025

  41. [49]

    Scaling up to excellence: Practicing model scaling for photo- realistic image restoration in the wild,

    F. Yu, J. Gu, Z. Li, J. Hu, X. Kong, X. Wang, J. He, Y . Qiao, and C. Dong, “Scaling up to excellence: Practicing model scaling for photo- realistic image restoration in the wild,” inCVPR, 2024

    2024

  42. [50]

    Low-light image enhancement via generative perceptual priors,

    H. Zhou, W. Dong, X. Liu, Y . Zhang, G. Zhai, and J. Chen, “Low-light image enhancement via generative perceptual priors,” inAAAI, 2025

    2025

  43. [51]

    Mul- tilingual E5 text embeddings: A technical report,

    L. Wang, N. Yang, X. Huang, L. Yang, R. Majumder, and F. Wei, “Mul- tilingual E5 text embeddings: A technical report,”arXiv:2402.05672, 2024

    2024 arXiv

  44. [52]

    L. J. Ba, J. R. Kiros, and G. E. Hinton,arXiv:1607.06450, 2016

    2016 arXiv

  45. [53]

    Structure- preserving deraining with residue channel prior guidance,

    Q. Yi, J. Li, Q. Dai, F. Fang, G. Zhang, and T. Zeng, “Structure- preserving deraining with residue channel prior guidance,” inICCV, 2021

    2021

  46. [54]

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

    S.-J. Cho, S.-W. Ji, J.-P. Hong, S.-W. Jung, and S.-J. Ko, “Rethinking coarse-to-fine approach in single image deblurring,” inICCV, 2021

    2021

  47. [55]

    Pearson correlation coefficient,

    I. Cohen, Y . Huang, J. Chen, J. Benesty, J. Benesty, J. Chen, Y . Huang, and I. Cohen, “Pearson correlation coefficient,”Noise reduction in speech processing, 2009

    2009

  48. [56]

    Desnownet: Context-aware deep network for snow removal,

    Y . Liu, D. Jaw, S. Huang, and J. Hwang, “Desnownet: Context-aware deep network for snow removal,”TIP, 2018

    2018

  49. [57]

    Heavy rain image restoration: Integrating physics model and conditional adversarial learning,

    R. Li, L. Cheong, and R. T. Tan, “Heavy rain image restoration: Integrating physics model and conditional adversarial learning,” in CVPR, 2019

    2019

  50. [58]

    Deep learning for seeing through window with raindrops,

    Y . Quan, S. Deng, Y . Chen, and H. Ji, “Deep learning for seeing through window with raindrops,” inICCV, 2019

    2019

  51. [59]

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

    T. Wang, X. Yang, K. Xu, S. Chen, Q. Zhang, and R. W. H. Lau, “Spatial attentive single-image deraining with a high quality real rain dataset,” inCVPR, 2019

    2019

  52. [60]

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

    W. Chen, H. Fang, J. Ding, C. Tsai, and S. Kuo, “JSTASR: joint size and transparency-aware snow removal algorithm based on modified partial convolution and veiling effect removal,” inECCV, 2020

    2020

  53. [61]

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

    X. Li, J. Wu, Z. Lin, H. Liu, and H. Zha, “Recurrent squeeze-and- excitation context aggregation net for single image deraining,” inECCV, 2018

    2018

  54. [62]

    Simple baselines for image restoration,

    L. Chen, X. Chu, X. Zhang, and J. Sun, “Simple baselines for image restoration,” inECCV, 2022

    2022

  55. [63]

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

    W. Chen, Z. Huang, C. Tsai, H. Yang, J. Ding, and S. Kuo, “Learning multiple adverse weather removal via two-stage knowledge learning and multi-contrastive regularization: Toward a unified model,” inCVPR, 2022

    2022

  56. [64]

    Learning weather-general and weather-specific features for image restoration under multiple adverse weather conditions,

    Y . Zhu, T. Wang, X. Fu, X. Yang, X. Guo, J. Dai, Y . Qiao, and X. Hu, “Learning weather-general and weather-specific features for image restoration under multiple adverse weather conditions,” inCVPR, 2023

    2023

  57. [65]

    Gridformer: Residual dense transformer with grid structure for image restoration in adverse weather conditions,

    T. Wang, K. Zhang, Z. Shao, W. Luo, B. Stenger, T. Lu, T. Kim, W. Liu, and H. Li, “Gridformer: Residual dense transformer with grid structure for image restoration in adverse weather conditions,”IJCV, 2024

    2024

  58. [66]

    Teaching tailored to talent: Adverse weather restoration via prompt pool and depth-anything constraint,

    S. Chen, T. Ye, K. Zhang, Z. Xing, Y . Lin, and L. Zhu, “Teaching tailored to talent: Adverse weather restoration via prompt pool and depth-anything constraint,” inECCV, 2024

    2024

  59. [67]

    Prompt to restore, restore to prompt: Cyclic prompting for universal adverse weather removal,

    R. Liao, F. Li, Y . Wei, Z. Shi, L. Zhang, H. Bai, and M. Wang, “Prompt to restore, restore to prompt: Cyclic prompting for universal adverse weather removal,”TIP, 2025

    2025

  60. [68]

    Unpaired image-to-image translation using cycle-consistent adversarial networks,

    J. Zhu, T. Park, P. Isola, and A. A. Efros, “Unpaired image-to-image translation using cycle-consistent adversarial networks,” inICCV, 2017

    2017

  61. [69]

    Image-to-image translation with conditional adversarial networks,

    P. Isola, J. Zhu, T. Zhou, and A. A. Efros, “Image-to-image translation with conditional adversarial networks,” inCVPR, 2017. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 15

    2017

  62. [70]

    Rain-free and residue hand-in-hand: A progressive coupled network for real-time image deraining,

    K. Jiang, Z. Wang, P. Yi, C. Chen, Z. Wang, X. Wang, J. Jiang, and C. Lin, “Rain-free and residue hand-in-hand: A progressive coupled network for real-time image deraining,”TIP, 2021

    2021

  63. [71]

    Multi-stage progressive image restoration,

    S. W. Zamir, A. Arora, S. H. Khan, M. Hayat, F. S. Khan, M. Yang, and L. Shao, “Multi-stage progressive image restoration,” inCVPR, 2021

    2021

  64. [72]

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

    X. Liu, M. Suganuma, Z. Sun, and T. Okatani, “Dual residual networks leveraging the potential of paired operations for image restoration,” in CVPR, 2019

    2019

  65. [73]

    Pre-trained image processing transformer,

    H. Chen, Y . Wang, T. Guo, C. Xu, Y . Deng, Z. Liu, S. Ma, C. Xu, C. Xu, and W. Gao, “Pre-trained image processing transformer,” inCVPR, 2021

    2021

  66. [74]

    MAXIM: multi-axis MLP for image processing,

    Z. Tu, H. Talebi, H. Zhang, F. Yang, P. Milanfar, A. C. Bovik, and Y . Li, “MAXIM: multi-axis MLP for image processing,” inCVPR, 2022

    2022

  67. [75]

    Scope of validity of psnr in im- age/video quality assessment,

    Q. Huynh-Thu and M. Ghanbari, “Scope of validity of psnr in im- age/video quality assessment,”Electronics letters, 2008

    2008

  68. [76]

    Image quality assessment: from error visibility to structural similarity,

    Z. Wang, A. C. Bovik, H. R. Sheikh, and E. P. Simoncelli, “Image quality assessment: from error visibility to structural similarity,”TIP, 2004

    2004

  69. [77]

    Adam: A method for stochastic optimization,

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” inICLR, 2015

    2015

  70. [78]

    SGDR: stochastic gradient descent with warm restarts,

    I. Loshchilov and F. Hutter, “SGDR: stochastic gradient descent with warm restarts,” inICLR, 2017

    2017

  71. [79]

    LDP: language- driven dual-pixel image defocus deblurring network,

    H. Yang, L. Pan, Y . Yang, R. I. Hartley, and M. Liu, “LDP: language- driven dual-pixel image defocus deblurring network,” inCVPR, 2024. Wanshu Fan(Member, IEEE) was born in Hei- longjiang, China. She received the Ph.D. degree in computational mathematics from Dalian Universit...

    2024

  72. [2016]

    His research interest includes image analysis and enhancement, and related vision problems

  73. [2019]

    Her research interests include deep learning and computer vision

    Now she is pursuing software engineering at Dalian University and is working hard to pursue a master’s degree. Her research interests include deep learning and computer vision. Liyan Wangis currently a Ph.D. student at the School of Mathematical Sciences, Dalian University of ...

    2018

Pith tools

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