Pith. sign in

REVIEW 5 major objections 6 minor 48 references

Unsupervised Network for Single Image Raindrop Removal

T0 review · 5 major / 6 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 establishes that an unsupervised cycle-consistent layer-separation network, guided by an iteratively refined transparency mask, outperforms all unpaired-image baselines for single-image raindrop removal on the NUS and RainDS…

desk verdict A plausible unsupervised raindrop-removal method with a solid ablation study, but its state-of-the-art claim is undercut by the omission of RainGAN, a directly comparable method cited in the paper itself. read the letter →

arxiv 2412.03019 v1 pith:WXSAZPTZ submitted 2024-12-04 cs.CV

Huijiao Wang , Shenghao Zhao , Lei Yu , Xulei Yang This is my paper · ORCID
classification cs.CV
keywords raindropremovalunsupervisedlearninglayerseparationcycle-consistentGANfeedbackmechanismiterativeneuralnetworktransparencymasksingleimagerestoration
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tackles single-image raindrop removal without requiring paired rainy and clean training images. It proposes a network that decomposes a rainy image into a clean background, a raindrop layer, and a transparency mask according to the linear blend model $I = (1-\alpha)\odot B + \alpha\odot R$. A cycle-consistent adversarial structure trained on unpaired image sets, combined with an iterative feedback loop that refines the mask across six shared-weight iterations, lets the model improve its separation progressively. On the NUS raindrop dataset and the RainDS real subset, the model reports the highest PSNR and SSIM among the unsupervised and weakly supervised baselines it compares against. The paper argues that unpaired training makes raindrop removal practical when paired real-world data is scarce.

What carries the argument

The load-bearing identity is the image-decomposition model $I = (1-\alpha)\odot B + \alpha\odot R$, which assumes every raindrop effect is a per-pixel linear blend of a clean background $B$ and a raindrop layer $R$ with a single transparency mask $\alpha$. Around this model, the machinery is a cycle-consistent GAN with one generator and two PatchGAN discriminators, trained with adversarial, cycle-consistency, identity, and sparsity losses. The distinctive mechanism is the iterative feedback network: $N$ sub-networks share weights, and the output mask $\alpha^{i-1}$ from iteration $i-1$ is concatenated with the input rainy image to feed iteration $i$, so the mask is refined from coarse to fine. Each iteration computes its own losses, later iterations are weighted more heavily by $K_i = i-1$, and the clean background from the last iteration is the final output.

What would settle it

Synthesize a raindrop as a pure refractive displacement of the background (no intensity change), feed it to the trained model, and compare the restored background against the original; if the displaced edge remains as a ghost and PSNR drops far below the reported roughly 27 dB, the single-alpha blend assumption fails.

Watch

Extended reading notes

Core claim

The discovery is that unpaired, cycle-consistent layer separation can beat existing translation-based unsupervised methods for raindrop removal when the separation is driven by a transparency mask that is refined iteratively. The generator splits $I$ into $B$, $R$, and $\alpha$; the recomposition $F(B,R,\alpha) = (1-\alpha)\odot B + \alpha\odot R$ must reproduce $I$ (cycle loss), while $B$ is pushed toward the clean-image distribution by a PatchGAN discriminator and an identity loss, and $\alpha$ is pushed toward zero (sparsity). Feeding the previous iteration's $\alpha$ back as extra input to the next of $N$ shared-parameter iterations yields coarse-to-fine masks. On the NUS dataset the model reaches $27.0562$ dB PSNR and $0.8738$ SSIM on test_a and $24.7124$ dB and $0.8281$ on test_b, surpassing the weakly supervised baseline by $1.5938$ and $1.4679$ dB respectively; on RainDS it reaches $21.7624$ dB and $0.7472$ SSIM, the best among the compared unpaired methods.

Load-bearing premise

The method assumes that a raindrop's visual effect can be undone by blending the clean background with a raindrop layer using a single per-pixel transparency mask, even though the paper notes that raindrops refract and defocus light from both foreground and background.

Editorial extensions

If this is right

  • Unpaired training removes the need for laboriously captured paired raindrop and clean photos, so the method can be re-targeted to a new camera or environment using only independent clean and rainy collections.
  • The learned transparency mask localizes the raindrops explicitly, giving a detection map as a by-product of restoration.
  • The feedback loop allows a fixed-weight network to be run for any number of refinement iterations, trading inference compute against restoration quality without retraining.
  • The decomposition into background, raindrop layer, and mask supports editing beyond restoration, such as removing individual drops or re-synthesizing rain.
  • On the NUS and RainDS benchmarks tested, the method reports higher PSNR and SSIM than all four unpaired baselines, with the largest margins on the harder test splits.

Reading between the lines

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

  • The same layer-separation plus feedback recipe could plausibly transfer to other adherent-occlusion problems — dust, mud, frost, or water on a lens — since none of the losses is specific to raindrop physics.
  • Because the linear blend model cannot represent refraction, the reported gains over translation baselines may shrink on images with large, strongly refractive drops; a dedicated test set of close-up droplet images would reveal whether a physics-aware decomposition is needed.
  • The sparsity loss encodes the prior that raindrops occupy a small fraction of pixels; on images with dense, overlapping droplets this prior could suppress legitimate mask values and cap performance.
  • A direct comparison with fully supervised methods on the same test splits would quantify how much accuracy is sacrificed for unpaired training, and could motivate a semi-supervised hybrid that uses a few paired samples alongside the unpaired collections.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 6 minor

Summary. The paper proposes an unsupervised single-image raindrop removal method that decomposes a rainy image into a clean background, a transparency mask, and a raindrop layer using the linear blend model of Eq. (1). The architecture combines a CycleGAN-style generator/discriminator setup with an iterative feedback mechanism that reuses the transparency mask from the previous iteration as additional input. The training losses are a weighted combination of GAN, cycle-consistency, identity, and sparsity terms. Experiments on the NUS and RainDS datasets report PSNR/SSIM gains over four unsupervised baselines, and an ablation study shows each loss and the iterative module contribute to the final performance. The paper's central claim is that this is a state-of-the-art unsupervised raindrop removal method, quantitatively surpassing the previous best by about 1.5 dB PSNR on NUS and achieving 21.76 dB on RainDS.

Significance. If the empirical results hold, the paper offers a practical unsupervised alternative to paired-data raindrop removal, with the feedback mechanism and sparsity-constrained decomposition being reasonable contributions. The ablation study is a useful step toward understanding the contribution of each loss. However, the significance is currently undercut by two load-bearing issues: (1) the identity loss in Eq. (6) is not an identity loss in any CycleGAN sense, because it compares the decomposed background of a rainy image to an unrelated clean image; and (2) the state-of-the-art claim is made without comparing against RainGAN, a directly applicable unsupervised raindrop removal method cited in the paper's own introduction. The degradation model of Eq. (1) is also a strong simplification that the paper does not critically assess.

major comments (5)
  1. [§3.3, Eq. (6)] The identity loss is mathematically and conceptually incorrect. The text states that 'sending B to G should still generate B' and therefore defines L_identity as E_B[||B_i - B||_1], where B_i is the generator output for a rainy input I and B is a random clean image from the training set. These are not corresponding images of the same scene, so the L1 penalty does not measure any identity property; it merely pulls the output toward arbitrary clean pixels. The ablation result in Table 3 for 'w/o L_identity' cannot be interpreted as validating an identity-preserving mechanism. This needs to be clarified: either the loss should be defined for the same input passed through the generator, or the paper should explain why an unpaired L1 term is a useful regularizer.
  2. [§4.2 and Tables 1-2] The claim of outperforming all unsupervised methods is unverified because RainGAN (Yan et al. [40], WACVW 2022), an unsupervised raindrop removal method based on decomposition and composition, is cited in Section 1 but omitted from the experimental comparisons. Section 4 states that 'the unsupervised model [21] is the only one we can find in the literature specifically designed for raindrop removal,' which is contradicted by the authors' own reference to [40]. Without quantitative results for RainGAN on NUS and RainDS, the state-of-the-art assertion is unsupported. The paper should either add this baseline or explicitly justify its exclusion.
  3. [§3.3 and Fig. 2] The method does not implement a cycle-consistency loss between two domains in the sense of CycleGAN. The 'cycle' here is a self-reconstruction: the generator decomposes the rainy input I and the reconstruction recombines the predicted layers to approximate I. Equation (5) is therefore a standard reconstruction loss, not a cycle-consistency loss. The terminology overstates the architectural novelty and may mislead readers about the relation to CycleGAN. The authors should describe this as a reconstruction loss and avoid the claim that they are using a cycle network.
  4. [§1 and Eq. (1)] The degradation model in Eq. (1) assumes every raindrop effect can be expressed as a per-pixel linear blend of a clean background B and a raindrop layer R with a single transparency mask α. The paper itself notes in Section 1 that raindrops 'alter the focus of both foreground and background, leading to unexpected blurring,' and that they cause background distortion. Refraction and defocus are not obviously captured by one alpha blend per pixel. While the benchmarks show good results, this modeling limitation should be acknowledged as a potential barrier to generalization; the current text presents Eq. (1) as if it were an exact physical model without discussing its validity.
  5. [§4.1, Eqs. (7) and (9)] The loss weighting is specified inconsistently. Equation (7) introduces K_i = i - 1 as a per-iteration weight for the GAN loss, but Section 4.1 then states that 'the hyper-parameters β1∼4 are 2 × 1.5^{i−1}, 10, 5, and 1,' implying an iteration-dependent β1 that does not appear in Eq. (9) where β1 is a single scalar. This makes the exact training objective ambiguous and hampers reproducibility. Please clarify whether the per-iteration weights are K_i, β1, or both, and give the exact final expression.
minor comments (6)
  1. [Abstract and §3.3] The abstract and Section 3.3 refer to a 'cycle network architecture' and 'cycle structure,' but the actual mechanism is a single-domain decomposition and reconstruction; consider using 'reconstruction' instead of 'cycle' to avoid confusion.
  2. [§3.4] The text contains a typo: 'Combing' should be 'Combining' in the sentence introducing the total loss.
  3. [§4.5] There are two minor language errors: 'the qualification of the generated clean images decays' should likely be 'quality,' and 'we can also obverse' should be 'observe.'
  4. [Fig. 6] The order of the methods in the figure caption (Input, UA_GAN, AGGAN, CycleGAN, Ours, GT) does not match the order described in the text of §4.2; please align them.
  5. [References] Reference [30] (Stollenga et al.) is missing the publication year and venue in the bibliography.
  6. [Implementation details] The paper states that the network structure is 'borrowed from CycleGAN' but does not provide the specific architecture of the generator (e.g., number of Dense ResNet blocks, channels). Adding these details would aid reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the method is an empirical architecture whose outputs are evaluated on held-out paired benchmarks, and the few self-citations are related-work context rather than load-bearing premises.

full rationale

The paper's derivation chain is not circular. The decomposition model in Eq. (1) posits I = (1-alpha)*B + alpha*R, and the network outputs B_i, R_i, alpha_i that are recomposed into the input via Eq. (3); the cycle loss in Eq. (5) is an auto-consistency constraint rather than a derivation of one target from another. The reported PSNR/SSIM values are computed on held-out paired test sets (NUS test_a/test_b and RainDS) against ground-truth background images, so no fitted parameter or training loss equals a reported test metric by construction. Hyperparameters beta_1..4 and K_i are training weights, and tuning them on validation/test data, while a generalization concern, is not a circularity. The identity loss in Eq. (6) is formally mis-specified and the paper's claim of training only on unpaired sets is weakened by the statement that 861 image pairs are used for training with 'input/truth data'; however, this is an experimental soundness and fairness issue, not a circular equation in the derivation. The self-citations ([34], [35], [38]) appear in related-work summaries and are not used to justify the core architecture, which is built on external CycleGAN [49], PatchGAN [8], and Double-DIP [11]. The omission of RainGAN [40] from quantitative comparisons is a completeness concern for the state-of-the-art claim, but it does not make the method's output definitionally dependent on its input. Overall, no step reduces a predicted quantity to a fitted input by construction.

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

The method rests on a handful of modeling assumptions rather than axioms in the mathematical sense. The free parameters are the loss weights and iteration count, tuned on the target benchmark. The invented latent factors (alpha and R) are unconstrained by ground truth and carry the main burden of the decomposition.

free parameters (2)
  • Loss weights beta_1 to beta_4 = beta_1 = 2 * 1.5^(i-1), beta_2 = 10, beta_3 = 5, beta_4 = 1
    Chosen by experiments on the benchmark datasets; the schedule weights later iterations more heavily and is part of the reported state-of-the-art numbers.
  • Iteration count N = 6
    Selected as the number of feedback iterations; the ablation shows IterNN helps, but no sweep over N is reported, so 6 is a hand-chosen design value.
assumptions (3)
  • domain assumption Linear alpha-composition model (Eq. 1) valid for real raindrops
    Assumes any rainy image can be written as (1-alpha)⊙B + alpha⊙R; real refraction and defocus may violate this linear blend.
  • domain assumption Unpaired image sets share content statistics
    CycleGAN-style training assumes rainy and clean piles have similar scenes; otherwise reconstruction and GAN losses can be satisfied by domain shifts that do not correspond to true removal.
  • domain assumption Raindrops are spatially sparse in the image
    The sparsity loss L1(alpha, 0) presupposes that most pixels are background, which may fail for dense rain or heavy lens contamination.
invented entities (2)
  • Transparency mask alpha
    purpose: Per-pixel map indicating raindrop locations, used to blend background and raindrop layers and to gate the feedback loop.
    A latent variable with no direct supervision; it is regularized only by the sparsity loss and reconstruction consistency.
  • Raindrop layer R
    purpose: Image containing raindrop effects, including droplets plus distortion and blur, separated from the background.
    A latent image with no direct supervision; because of Eq. (1), many pairs (alpha, R) can explain the same input, so R is not uniquely identifiable without extra priors.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Unsupervised Network for Single Image Raindrop Removal." pith.science (2026). https://pith.science/paper/WXSAZPTZ

@misc{pith2026241203019,
  author       = {Pith},
  title        = {Pith review of: Unsupervised Network for Single Image Raindrop Removal},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WXSAZPTZ}},
  note         = {Machine review of arXiv:2412.03019}
}
read the original abstract

Image quality degradation caused by raindrops is one of the most important but challenging problems that reduce the performance of vision systems. Most existing raindrop removal algorithms are based on a supervised learning method using pairwise images, which are hard to obtain in real-world applications. This study proposes a deep neural network for raindrop removal based on unsupervised learning, which only requires two unpaired image sets with and without raindrops. Our proposed model performs layer separation based on cycle network architecture, which aims to separate a rainy image into a raindrop layer, a transparency mask, and a clean background layer. The clean background layer is the target raindrop removal result, while the transparency mask indicates the spatial locations of the raindrops. In addition, the proposed model applies a feedback mechanism to benefit layer separation by refining low-level representation with high-level information. i.e., the output of the previous iteration is used as input for the next iteration, together with the input image with raindrops. As a result, raindrops could be gradually removed through this feedback manner. Extensive experiments on raindrop benchmark datasets demonstrate the effectiveness of the proposed method on quantitative metrics and visual quality.

Figures

Figures reproduced from arXiv: 2412.03019 by the authors.

Figure 1
Figure 1. Examples of rainy images in the datasets we used. (a) and (b) are from the NUS raindrop dataset [25]. (c) and (d) are from RainDS [26]. In response to the challenges posed by real-world single-image raindrop removal, this paper presents a pioneering unsupervised image decomposition method coupled with a feedback network. Diverging from conventional supervised learning paradigms, our innovative approach relies solely… view at source ↗
Figure 2
Figure 2. Overall flowchart of our proposed method [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Diagram of Image Decomposition. A heat map represents the transparency mask. The closer the value of a pixel is to 1, the closer its color is to red. Meanwhile, we introduce two discriminators individually to determine whether the split clean background and the reconstructed rainy image conform to the true distribution. The clean split background is what we need in the end. The whole process does not require pairwis… view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Detailed flowchart of the iterative network as the generator in [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Samples of rainy images, transparency mask, and clean background images at different iterations in our iterative neural network. single-image raindrop removal. Meanwhile, we can also reassemble the decomposed images into the original rainy images. 3.2. Iterative Neural…
Figure 6
Figure 6. Figure 6: Results on NUS Raindrop dataset. From top to bottom: the input raindrop image, UA_GAN, AGGAN, CycleGAN, our model, and the ground truth. 4.2. Compared with the State-of-the-Arts on NUS Raindrop Dataset NUS raindrop dataset [25] has been widely used in recent years. The…
Figure 7
Figure 7. Figure 7: Results on RainDS dataset. From left to right: the input raindrop image, the ground truth, UA_GAN, AGGAN, CycleGAN, and ours. model [21] is the most recent model designed for raindrop removal training on unpaired datasets. Our method surpasses the unsupervised model by…
Figure 8
Figure 8. Figure 8: Visualization of intermediate output from feedback mechanism on NUS Raindrop dataset and RainDS dataset. From top to bottom: the input raindrop image, the learned transparency mask from iteration 𝑖 from 1 to 𝑁. the similar color and texture details as the ground truth.…
Figure 9
Figure 9. Figure 9: Component analysis. Results of the proposed model and its variants. raindrop benchmark datasets show the effectiveness of the proposed method. We hope this work can motivate further research on designing efficient network structures for unsupervised raindrop removal. C…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

48 extracted references · 30 canonical work pages

  1. [21]

    Weakly supervised learning for raindrop removal on a single image

    Luo, W., Lai, J., Xie, X., 2021. Weakly supervised learning for raindrop removal on a single image. IEEE Transactions on Circuits and Systems for Video Technology 31, 1673–1683. doi:10.1109/TCSVT.2020.3014267

    arXiv 2021

  2. [40]

    Raingan: Unsupervised raindrop removal via decomposition and composition, in: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) Workshops, pp

    Yan, X., Loke, Y.R., 2022. Raingan: Unsupervised raindrop removal via decomposition and composition, in: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) Workshops, pp. 14–23. doi:10.1109/WACVW54805.2022.00007

    arXiv 2022

  3. [1]

    Unsupervised attention-guided image-to-image translation, in: Proceedings of the 32nd International Conference on Neural Information Processing Systems, p

    Alami Mejjati, Y., Richardt, C., Tompkin, J., Cosker, D., Kim, K.I., 2018. Unsupervised attention-guided image-to-image translation, in: Proceedings of the 32nd International Conference on Neural Information Processing Systems, p. 3697–3707. doi:10.5555/3327144. 3327286

    work page doi:10.5555/3327144 2018

  4. [2]

    Analysisofrainandsnowinfrequencyspace

    Barnum,P.C.,Narasimhan,S.,Kanade,T.,2010. Analysisofrainandsnowinfrequencyspace. InternationalJournalofComputerVision86, 256–274. doi:10.1007/s11263-008-0200-2

    work page doi:10.1007/s11263-008-0200-2 2010

  5. [3]

    Non-localimagedehazing,in:IEEEConferenceonComputerVisionandPatternRecognition,pp

    Berman,D.,Treibitz,T.,Avidan,S.,2016. Non-localimagedehazing,in:IEEEConferenceonComputerVisionandPatternRecognition,pp. 1674–1682. doi:10.1109/CVPR.2016.185

    work page doi:10.1109/cvpr.2016.185 2016

  6. [4]

    Rain or snow detection in image sequences through use of a histogram of orientation of streaks

    Bossu, J., Hautiere, N., Tarel, J.P., 2011. Rain or snow detection in image sequences through use of a histogram of orientation of streaks. International Journal of Computer Vision 93, 348–367. doi:10.1007/s11263-011-0421-7

    work page doi:10.1007/s11263-011-0421-7 2011

  7. [5]

    Brewer, N., Liu, N., 2008. Using the shape characteristics of rain to identify and remove rain from video, in: Proceedings of the Joint IAPR International Workshops on Structural and Syntactic Pattern Recognition, pp. 451–458. doi:10.1007/978-3-540-89689-0_49

    work page doi:10.1007/978-3-540-89689-0_49 2008

  8. [6]

    Unsupervisedderaining:Whereasymmetriccontrastivelearning meets self-similarity

    Chang,Y.,Guo,Y.,Ye,Y.,Yu,C.,Zhu,L.,Zhao,X.,Yan,L.,Tian,Y.,2023. Unsupervisedderaining:Whereasymmetriccontrastivelearning meets self-similarity. IEEE Transactions on Pattern Analysis and Machine Intelligence , 1–18doi:10.1109/TPAMI.2023.3321311

    work page arXiv 2023

Show all 48 references
  1. [7]

    On the effectiveness of visible watermarks, in: IEEE Conference on Computer Vision and Pattern Recognition, pp

    Dekel, T., Rubinstein, M., Liu, C., Freeman, W., 2017. On the effectiveness of visible watermarks, in: IEEE Conference on Computer Vision and Pattern Recognition, pp. 6864–6872. doi:10.1109/CVPR.2017.726

    2017 doi

  2. [8]

    Patch-based image inpainting with generative adversarial networks

    Demir, U., Unal, G., 2018. Patch-based image inpainting with generative adversarial networks. arXiv preprint arXiv:1803.07422

    2018 arXiv

  3. [9]

    Restoring an image taken through a window covered with dirt or rain, in: IEEE International Conference on Computer Vision, pp

    Eigen, D., Krishnan, D., Fergus, R., 2013. Restoring an image taken through a window covered with dirt or rain, in: IEEE International Conference on Computer Vision, pp. 633–640. doi:10.1109/ICCV.2013.84

    2013 doi

  4. [10]

    Co-segmentation by composition, in: IEEE International Conference on Computer Vision, pp

    Faktor, A., Irani, M., 2013. Co-segmentation by composition, in: IEEE International Conference on Computer Vision, pp. 1297–1304. doi:10.1109/ICCV.2013.164

    2013 doi

  5. [11]

    Double-DIP

    Gandelsman, Y., Shocher, A., Irani, M., 2019. “Double-DIP”: Unsupervised image decomposition via coupled deep-image-priors, in: IEEE Conference on Computer Vision and Pattern Recognition, pp. 11018–11027. doi:10.1109/CVPR.2019.01128

    2019

  6. [12]

    Detection and removal of rain from videos, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp

    Garg, K., Nayar, S.K., 2004. Detection and removal of rain from videos, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. I–528. doi:10.1109/CVPR.2004.1315077

    2004 arXiv

  7. [13]

    When does a camera see rain?, in: Proceedings of the IEEE International Conference on Computer Vision, pp

    Garg, K., Nayar, S.K., 2005. When does a camera see rain?, in: Proceedings of the IEEE International Conference on Computer Vision, pp. 1067–1074. doi:10.1109/ICCV.2005.253

    2005 doi

  8. [14]

    Photorealisticrenderingofrainstreaks

    Garg,K.,Nayar,S.K.,2006. Photorealisticrenderingofrainstreaks. ACMTransactionsonGraphics25,996–1002. doi: 10.1145/1141911. 1141985

    2006 doi

  9. [15]

    Vision and rain

    Garg, K., Nayar, S.K., 2007. Vision and rain. International Journal of Computer Vision 75, 3–27. doi:10.1007/s11263-006-0028-6

    2007 doi

  10. [16]

    Learning from synthetic photorealistic raindrop for single image raindrop removal, in: IEEE International Conference on Computer Vision Workshop, pp

    Hao, Z., You, S., Li, Y., Li, K., Lu, F., 2019. Learning from synthetic photorealistic raindrop for single image raindrop removal, in: IEEE International Conference on Computer Vision Workshop, pp. 4340–4349. doi:10.1109/ICCVW.2019.00534

    2019

  11. [17]

    Adherent mist and raindrop removal from a single image using attentive convolutional network

    He, D., Shang, X., Luo, J., 2022. Adherent mist and raindrop removal from a single image using attentive convolutional network. Neurocomputing 505, 178–187. doi:10.1016/j.neucom.2022.07.032

    2022 doi

  12. [18]

    Single-image deraining using an adaptive nonlocal means filter, in: IEEE International Conference on Image Processing, pp

    Kim, J.H., Lee, C., Sim, J.Y., Kim, C.S., 2013. Single-image deraining using an adaptive nonlocal means filter, in: IEEE International Conference on Image Processing, pp. 914–917. doi:10.1109/ICIP.2013.6738189

    2013

  13. [19]

    All in one bad weather removal using architectural search, in: IEEE Conference on Computer Vision and Pattern Recognition, pp

    Li, R., Tan, R.T., Cheong, L.F., 2020. All in one bad weather removal using architectural search, in: IEEE Conference on Computer Vision and Pattern Recognition, pp. 3172–3182. doi:10.1109/CVPR42600.2020.00324

    2020

  14. [20]

    Feedback network for image super-resolution, in: IEEE Conference on Computer Vision and Pattern Recognition, pp

    Li, Z., Yang, J., Liu, Z., Yang, X., Jeon, G., Wu, W., 2019. Feedback network for image super-resolution, in: IEEE Conference on Computer Vision and Pattern Recognition, pp. 3867–3876. doi:10.1109/CVPR.2019.00399

    2019

  15. [22]

    Weakly-supervised action localization with background modeling, in: IEEE International Conference on Computer Vision, pp

    Nguyen, P., Ramanan, D., Fowlkes, C., 2019. Weakly-supervised action localization with background modeling, in: IEEE International Conference on Computer Vision, pp. 5501–5510. doi:10.1109/ICCV.2019.00560

    2019

  16. [23]

    Acomparisonstudyofcreditcardfrauddetection:Supervisedversusunsupervised

    Niu,X.,Wang,L.,Yang,X.,2019. Acomparisonstudyofcreditcardfrauddetection:Supervisedversusunsupervised. CoRRabs/1904.10604. URL: http://arxiv.org/abs/1904.10604, arXiv:1904.10604

    2019 arXiv

  17. [24]

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

    Özdenizci, O., Legenstein, R., 2023. Restoring vision in adverse weather conditions with patch-based denoising diffusion models. IEEE Transactions on Pattern Analysis and Machine Intelligence , 1–12doi:10.1109/TPAMI.2023.3238179

    2023

  18. [25]

    Attentive generative adversarial network for raindrop removal from a single image, in: IEEE Conference on Computer Vision and Pattern Recognition, pp

    Qian, R., Tan, R.T., Yang, W., Su, J., Liu, J., 2018. Attentive generative adversarial network for raindrop removal from a single image, in: IEEE Conference on Computer Vision and Pattern Recognition, pp. 2482–2491. doi:10.1109/CVPR.2018.00263

    2018

  19. [26]

    Removing raindrops and rain streaks in one go, in: IEEE Conference on Computer Vision and Pattern Recognition, pp

    Quan, R., Yu, X., Liang, Y., Yang, Y., 2021. Removing raindrops and rain streaks in one go, in: IEEE Conference on Computer Vision and Pattern Recognition, pp. 9147–9156. doi:10.1109/CVPR46437.2021.00903

    2021

  20. [27]

    Deep learning for seeing through window with raindrops, in: IEEE International Conference on Computer Vision, pp

    Quan, Y., Deng, S., Chen, Y., Ji, H., 2019. Deep learning for seeing through window with raindrops, in: IEEE International Conference on Computer Vision, pp. 2463–2471. doi:10.1109/ICCV.2019.00255

    2019

  21. [28]

    Video-basedraindropdetectionforimprovedimageregistration,in:IEEEInternationalConferenceonComputer Vision Workshops, pp

    Roser,M.,Geiger,A.,2009. Video-basedraindropdetectionforimprovedimageregistration,in:IEEEInternationalConferenceonComputer Vision Workshops, pp. 570–577. doi:10.1109/ICCVW.2009.5457650

    2009

  22. [29]

    Realistic modeling of water droplets for monocular adherent raindrop recognition using bezier curves, in: Asian Conference on Computer Vision, pp

    Roser, M., Kurz, J., Geiger, A., 2010. Realistic modeling of water droplets for monocular adherent raindrop recognition using bezier curves, in: Asian Conference on Computer Vision, pp. 235–244. doi:10.1007/978-3-642-22819-3_24

    2010 doi

  23. [30]

    Deep networks with internal selective attention through feedback connections, in: Proceedings of the 27th International Conference on Neural Information Processing Systems, p

    Stollenga, M.F., Masci, J., Gomez, F., Schmidhuber, J., . Deep networks with internal selective attention through feedback connections, in: Proceedings of the 27th International Conference on Neural Information Processing Systems, p. 3545–3553. doi:10.5555/2969033. 2969222

    doi

  24. [31]

    Attention-guided generative adversarial networks for unsupervised image-to-image translation, in: International Joint Conference on Neural Networks, IEEE

    Tang, H., Xu, D., Sebe, N., Yan, Y., 2019. Attention-guided generative adversarial networks for unsupervised image-to-image translation, in: International Joint Conference on Neural Networks, IEEE. pp. 1–8. doi:10.1109/IJCNN.2019.8851881. H. Wang et al.:Preprint submitted to E...

    2019

  25. [32]

    Semi-supervised and unsupervised methods for heart sounds classification in restricted data environments

    Unnikrishnan, B., Singh, P.R., Yang, X., Chua, M.C.H., 2020. Semi-supervised and unsupervised methods for heart sounds classification in restricted data environments. CoRR abs/2006.02610. URL:https://arxiv.org/abs/2006.02610, arXiv:2006.02610

    2020 arXiv

  26. [34]

    Facial feature embedded cyclegan for vis-nir translation, in: ICASSP 2020 - 2020 IEEEInternationalConferenceonAcoustics,SpeechandSignalProcessing(ICASSP),pp.1903–1907

    Wang, H., Zhang, H., Yu, L., Wang, L., Yang, X., 2020. Facial feature embedded cyclegan for vis-nir translation, in: ICASSP 2020 - 2020 IEEEInternationalConferenceonAcoustics,SpeechandSignalProcessing(ICASSP),pp.1903–1907. doi: 10.1109/ICASSP40776.2020. 9054007

    2020

  27. [35]

    Facialfeatureembeddedcycleganforvis–nirtranslation

    Wang,H.,Zhang,H.,Yu,L.,Yang,X.,2023. Facialfeatureembeddedcycleganforvis–nirtranslation. MultidimensionalSyst.SignalProcess. 34, 423–446. URL:https://doi.org/10.1007/s11045-023-00871-1, doi:10.1007/s11045-023-00871-1

    2023 doi

  28. [36]

    Raindropremovalfromasingleimageusingatwo-stepgenerativeadversarialnetwork

    Xia,H.,Lan,Y.,Song,S.,Li,H.,2022. Raindropremovalfromasingleimageusingatwo-stepgenerativeadversarialnetwork. Signal,Image and Video Processing 16, 677–684. doi:10.1007/s11760-021-02007-z

    2022 doi

  29. [37]

    An improved guidance image based method to remove rain and snow in a single image

    Xu, J., Zhao, W., Liu, P., Tang, X., 2012. An improved guidance image based method to remove rain and snow in a single image. Computer and Information Science 5, 49. doi:10.5539/cis.v5n3p49

    2012 doi

  30. [38]

    Iterative contrastive learning for single image raindrop removal, in: 2022 IEEE International Conference on Image Processing (ICIP), pp

    Xulei, Y., Peisheng, Q., Li, W., Shenghao, Z., Cen, C., Xiaoli, L., Zeng, Z., 2022. Iterative contrastive learning for single image raindrop removal, in: 2022 IEEE International Conference on Image Processing (ICIP), pp. 456–460. doi:10.1109/ICIP46576.2022.9897979

    2022

  31. [39]

    Feature-aligned video raindrop removal with temporal constraints

    Yan, W., Xu, L., Yang, W., Tan, R., 2022. Feature-aligned video raindrop removal with temporal constraints. IEEE Transactions on Image Processing 31, 3440–3448. doi:10.48550/arXiv.2205.14574

    2022 doi arXiv

  32. [41]

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

    Yang, W., Tan, R.T., Feng, J., Guo, Z., Yan, S., Liu, J., 2020. Joint rain detection and removal from a single image with contextualized deep networks. IEEE Transactions on Pattern Analysis and Machine Intelligence 42, 1377–1393. doi:10.1109/TPAMI.2019.2895793

    2020

  33. [42]

    Single image deraining: From model-based to data-driven and beyond

    Yang, W., Tan, R.T., Wang, S., Fang, Y., Liu, J., 2021. Single image deraining: From model-based to data-driven and beyond. IEEE Transactions on Pattern Analysis and Machine Intelligence 43, 4059–4077. doi:10.1109/TPAMI.2020.2995190

    2021

  34. [44]

    Adherent raindrop modeling, detectionand removal in video

    You, S., Tan, R.T., Kawakami, R., Mukaigawa, Y., Ikeuchi, K., 2015. Adherent raindrop modeling, detectionand removal in video. IEEE transactions on pattern analysis and machine intelligence 38, 1721–1733. doi:10.1109/TPAMI.2015.2491937

    2015

  35. [45]

    Single image deraining with continuous rain density estimation

    Yu, L., Wang, B., He, J., Xia, G.S., Yang, W., 2023. Single image deraining with continuous rain density estimation. IEEE Transactions on Multimedia 25, 443–456. doi:10.1109/TMM.2021.3127360

    2023

  36. [46]

    Yu, Y., Yang, W., Tan, Y.P., Kot, A.C., 2022. Towards robust rain removal against adversarial attacks: A comprehensive benchmark analysis and beyond, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 6013–6022. doi:10.1109/CVPR52...

    2022

  37. [47]

    Feedback networks, in: 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp

    Zamir, A.R., Wu, T.L., Sun, L., Shen, W.B., Shi, B.E., Malik, J., Savarese, S., 2017. Feedback networks, in: 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 1808–1817. doi:10.1109/CVPR.2017.196

    2017 doi

  38. [48]

    Rainremovalinvideobycombiningtemporalandchromaticproperties,in:Proceedings of the IEEE International Conference on Multimedia and Expo, pp

    Zhang,X.,Li,H.,Qi,Y.,Leow,W.K.,Ng,T.K.,2006. Rainremovalinvideobycombiningtemporalandchromaticproperties,in:Proceedings of the IEEE International Conference on Multimedia and Expo, pp. 461–464. doi:10.1109/ICME.2006.262572

    2006

  39. [49]

    Unpaired image-to-image translation using cycle-consistent adversarial networks, in: IEEE International Conference on Computer Vision, pp

    Zhu, J., Park, T., Isola, P., Efros, A.A., 2017. Unpaired image-to-image translation using cycle-consistent adversarial networks, in: IEEE International Conference on Computer Vision, pp. 2242–2251. doi:10.1109/ICCV.2017.244

    2017 doi

  40. [50]

    Laplacianencoder-decodernetworkforraindropremoval

    Zini,S.,Buzzelli,M.,2022. Laplacianencoder-decodernetworkforraindropremoval. PatternRecognitionLetters158,24–33. doi: 10.1016/ j.patrec.2022.04.016. H. Wang et al.:Preprint submitted to Elsevier Page 14 of 14

    2022

Pith tools

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