Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

DRD-Net: Detail-recovery Image Deraining via Context Aggregation Networks

T0 review · 4 major / 5 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 A two-stage network that first removes rain streaks and then repairs the details those removals blur can outperform existing single-image deraining methods.

desk verdict DRD-Net's detail-repair idea is plausible and the ablation supports it, but duplicated baseline numbers make the state-of-the-art claim unreliable until corrected. read the letter →

arxiv 1908.10267 v2 pith:6BXYGHZD submitted 2019-08-27 eess.IV cs.CV

Sen Deng , Mingqiang Wei , Jun Wang , Luming Liang , Haoran Xie , Meng Wang This is my paper · ORCID
classification eess.IVcs.CV
keywords singleimagederainingdetailrecoverysqueeze-and-excitationdilatedconvolutioncontextaggregationresiduallearningrainstreakremovalrestoration
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to show that the detail loss caused by single-image deraining is not an unavoidable side effect: it can be explicitly reversed by a second subnetwork dedicated to detail repair. The proposed DRD-Net first learns to predict rain streaks R from a rainy image O and subtracts them, then a detail repair network built from dilated-context blocks adds back the fine structure that the subtraction blurred. On the three synthetic benchmarks Rain200L, Rain200H, and Rain800, the full model reports higher PSNR and SSIM than six earlier methods, and the authors show examples where fine details such as zebra stripes stay intact. If correct, this reframes deraining as a two-step inverse problem rather than a single mapping, and suggests that explicit detail recovery is worth adding to other image restoration tasks.

What carries the argument

Two building blocks carry the argument. The rain residual network uses rain residual blocks: a standard residual block wrapped in a squeeze-and-excitation (SE) operation, which learns a per-channel weight and lets the network emphasize feature channels carrying spatial context; the paper visualizes that the highest-weighted feature maps are indeed the context-rich ones. The detail repair network stacks 16 structure detail context aggregation blocks (SDCABs): each SDCAB runs three dilated convolutions with dilation rates 1, 3, and 5 on the same feature map, concatenates their outputs, compresses with a 1×1 convolution, and adds the input back. These dilated rates give the stack a receptive field of 231×231 at depth 35, which the paper argues is large enough to gather the context needed to reconstruct details rather than hallucinate them.

What would settle it

Run DRD-Net on a paired real-world rainy/clean benchmark or a physically based rain renderer that breaks O = B + R with scattering and motion-blurred streaks; if its PSNR/SSIM advantage over earlier methods disappears, or the detail-repair subnetwork outputs near zero when given a clean image (i.e., it hallucinates rather than recovers actual lost detail), the paper's central claim is falsified.

Watch

Extended reading notes

Core claim

The central claim is that a derained image has lost real, recoverable information, and that a network can be trained to predict precisely that lost information. Formally, the rainy observation is modeled as O = B + R; the rain residual network learns a function f with f(O) ≈ R, giving a preliminary derained image I_p = O - R. The detail repair network then learns a function g with I_p + g(O) ≈ B, so g(O) is the network's estimate of the detail component destroyed by removing R. The two losses are combined as λ1 ||f(O) - R||² + λ2 ||I_p + g(O) - B||² with λ1 = 0.1 and λ2 = 1.0. The authors argue that the second subnetwork, not just a deeper first network, is what lets the method preserve small high-frequency structures, and they support this with ablations showing the full two-subnetwork model beats the rain residual network alone on all three datasets.

Load-bearing premise

The whole pipeline assumes real rain is well approximated by the synthetic additive model O = B + R, with rain streaks separable from background details, so that subtracting the predicted R from O leaves a detail-lost but recoverable image; if real rain violates this assumption, the two subnetworks may simply be fitting synthetic statistics.

Editorial extensions

If this is right

  • If the central claim holds, deraining systems should be built as two stages rather than one, with the second stage explicitly trained on the residual between the derained image and the ground truth.
  • The reported gains are largest on the heavy-rain set Rain200H, so the detail-repair mechanism matters most when rain and background structure overlap in the high-frequency band.
  • The ablation results imply that each added component—SE weighting, a residual detail network, and the dilated-context block—pays off monotonically on all three benchmarks.
  • The paper's downstream experiment suggests that removing rain this way lowers a commercial vision recognizer's confidence that an image contains rain, which is a direct practical benefit if confirmed.

Reading between the lines

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

  • Our inference: the two-stage remove-then-repair scheme is not specific to rain; the same SDCAB-based repair network could be attached to any restoration front end that blurs high-frequency structure, such as dehazing or denoising, and the ablation logic suggests it would help there too.
  • Our inference: the paper's difference image Y-X implies the detail network acts as a high-frequency corrector, so one testable prediction is that the energy of g(O) concentrates along edges, which could be verified by gradient-magnitude statistics.
  • Our inference: because the architecture depends on the additive model O = B + R, the method's real-world advantage is not guaranteed by these experiments; a fair test would use paired real rain/clean images or a physically based renderer with scattering and motion-blurred streaks.
  • Our inference: if real rain follows the atmospheric-scattering variant O = α(B + ΣR_i) + (1−α)A, the detail repair network may need the transmission α as an additional input to correctly recover details in foggy heavy-rain scenes.
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

4 major / 5 minor

Summary. The paper proposes DRD-Net, an end-to-end single-image deraining network composed of two sub-networks: a rain residual network that predicts rain streaks using squeeze-and-excitation residual blocks, and a detail repair network built from a newly proposed structure detail context aggregation block (SDCAB) with dilated convolutions, intended to recover image details lost during deraining. The two sub-networks are trained jointly with a weighted L2 loss. The method is evaluated on Rain200L, Rain200H, and Rain800 synthetic benchmarks plus real-world images, with PSNR/SSIM comparisons against GMM, DSC, DDN, RESCAN, DAF-Net, and UGSM, and the paper reports state-of-the-art quantitative results and qualitative detail preservation.

Significance. The proposed architecture is a clear and plausible contribution to the deraining literature: the two-stage design with a dedicated detail-repair network is well motivated, and the ablation study in Table 3 supports the contribution of the SDCAB block. The authors also provide source code on GitHub, which aids reproducibility. However, the central quantitative claim currently rests on a table that contains apparent transcription errors, and the specification of the detail-repair network input is internally inconsistent. If the quantitative results survive verification and the architecture description is corrected, the paper would be a useful contribution to the community.

major comments (4)
  1. [Section 5.2, Table 2 and Fig. 1] Table 2 reports identical PSNR/SSIM values for DDN and GMM on Rain800 (24.04/0.8675), and Fig. 1 lists the same PSNR/SSIM for GMM and DSC (10.61/0.4675). Two structurally unrelated baselines producing bit-identical metrics on the same data is effectively impossible unless the values were copied or transcribed incorrectly. Since the reported margin over RESCAN on Rain200L is only 0.08 dB (37.15 vs. 37.07), even a small correction to a baseline entry could change the ranking. The authors must verify every baseline number, either by rerunning with the provided code under identical protocols or by clearly citing the original source of each number, and report whether the values are reproduced or taken from previous papers.
  2. [Section 4.2, Eqs. (6) and (10)] There is an internal inconsistency about the input to the detail repair network. The text says the network operates on the preliminarily derained image Ip, but Eq. (6) trains a function g(·) on the rainy image Oi, and Eq. (10) sets layer0 = PRelu(Conv 3x3(O)), i.e., the network input is the rainy image. If the network takes the rainy image as input, the surrounding description should be corrected; if it takes Ip, then Eqs. (6) and (10) are wrong. This ambiguity materially affects reproducibility of the proposed method and needs to be resolved.
  3. [Section 4.2 and Table 1] The dilation structure of the SDCAB is described inconsistently. In Section 4.2, SDCAB is said to contain dilated convolutions with dilation scales 1, 3, and 5, and Eq. (7) uses Conv 3x3,d1, Conv 3x3,d3, and Conv 3x3,d5. However, Table 1 lists a dilation value of 7 for layers 1 through 16. These two descriptions cannot both be correct. The authors should clarify the exact dilation rates used and re-derive the receptive field row accordingly.
  4. [Section 5.2, Table 2] The paper states that all methods were 'performed in the same training and testing datasets for fair comparison,' but does not specify whether baseline models were retrained by the authors or evaluated using pretrained models from their original releases. Given the duplicate baseline values in Table 2, the provenance of every baseline number needs to be stated explicitly. Without this information, the fair-comparison claim and the resulting state-of-the-art claim are not fully supported.
minor comments (5)
  1. [Section 5.1] The sentence 'three two synthetic datasets' contains a typo and should read 'three synthetic datasets.'
  2. [Section 5.1 and Table 4] Minor typographical issues include 'min-batch' (should be 'mini-batch'), 'Table. 4' (should be 'Table 4'), and 'Tabel 1' in the caption of Table 1.
  3. [Section 5.2] The text says the comparison includes UGSM [29], but UGSM does not appear in Table 2; it appears only in the real-world qualitative figures. The authors should either add UGSM to the quantitative table or remove it from the list of compared methods.
  4. [Fig. 1] The caption of Fig. 1 is duplicated in the manuscript text; please remove the repeated caption block.
  5. [Throughout] The phrase 'reception field' is used in several places; the standard term is 'receptive field.'

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: DRD-Net is a supervised deraining system evaluated on held-out test data; its loss functions do not reduce to its own predictions, and the only overlapping-author citation is a non-load-bearing related-work pointer.

full rationale

The paper's derivation chain is a supervised learning pipeline, not a first-principles reduction. Eq. (3) trains the rain residual network f against the synthetic ground-truth rain layer Rhat; Eq. (6) trains the detail repair network g against the ground-truth clean image Ihat; Eq. (11) combines both with fixed weights lambda1=0.1 and lambda2=1.0. The final output Ic (Eq. 10) is computed from network outputs, and performance is measured on held-out portions of Rain200L/Rain200H/Rain800 and on real-world images (Figs. 6-15, Table 2). No fitted parameter is relabeled as a prediction: the loss weights are fixed in advance, and the test-set PSNR/SSIM values are not used to define the objective. The ablation study in Table 3 and the SE/SDCAB analyses are internal empirical checks, not circular arguments. The only citation with overlapping authors is [6] (Guo, Xie, Liu, Wei, Wang), used as a related-work example of low-rank priors; it is not load-bearing, and no uniqueness theorem or ansatz is imported from the authors' prior work. I note one non-circular data-integrity concern: Table 2 lists DDN and GMM identically on Rain800 (24.04/0.8675), and Fig. 1 lists GMM and DSC identically (10.61/0.4675); these duplicates undermine the reliability of the quantitative comparison, but they are a correctness/reproducibility issue, not a circularity of the claimed derivation. The stated limitation (more parameters and training time) is also not circular. Overall, the central claim that DRD-Net outperforms baselines is an external empirical claim, self-contained against benchmarks, so the circularity score is 0.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

The paper introduces a new architectural block (SDCAB) but no new physical or conceptual entities. The central claim rests on hand-chosen hyperparameters and on the standard additive rain model, which is inherited from prior work.

free parameters (4)
  • lambda1 and lambda2 loss weights = 0.1 and 1.0
    In Eq. (11), these weights balance the rain residual loss and the detail repair loss. They are chosen by hand and affect the final performance.
  • Dilation rates in SDCAB = 1, 3, 5
    In Eq. (7), the three dilation rates define the receptive field of the SDCAB. They are manually selected, not derived.
  • Network depth and width = Depth 16+3, feature maps M=64 (best setting in Table 5)
    The number of SDCAB/RRB blocks and feature maps are chosen by ablation study, not by a principled derivation.
  • Learning rate schedule = 0.01, halved every 15 epochs
    A standard heuristic for Adam optimization; the specific schedule is chosen by hand.
assumptions (3)
  • domain assumption The rainy image is the sum of the clean background and rain streaks: O = B + R, or the heavy-rain variant O = alpha(B + sum Ri) + (1-alpha)A.
    This model forms the basis of the rain residual network and is cited from prior work. If real rain is not additive, the subtraction step Ip = O - R is invalid.
  • domain assumption Rain streaks and image details are separable high-frequency components that a learned residual mapping can distinguish.
    The entire architecture treats rain prediction and detail recovery as two separately learnable tasks. The paper does not prove this separability; it assumes it.
  • domain assumption The synthetic datasets Rain200L, Rain200H and Rain800 are valid proxies for real-world deraining performance.
    Quantitative claims are based on these synthetic test sets, and the real-world validation is only qualitative.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DRD-Net: Detail-recovery Image Deraining via Context Aggregation Networks." pith.science (2026). https://pith.science/paper/6BXYGHZD

@misc{pith2026190810267,
  author       = {Pith},
  title        = {Pith review of: DRD-Net: Detail-recovery Image Deraining via Context Aggregation Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6BXYGHZD}},
  note         = {Machine review of arXiv:1908.10267}
}
read the original abstract

Image deraining is a fundamental, yet not well-solved problem in computer vision and graphics. The traditional image deraining approaches commonly behave ineffectively in medium and heavy rain removal, while the learning-based ones lead to image degradations such as the loss of image details, halo artifacts and/or color distortion. Unlike existing image deraining approaches that lack the detail-recovery mechanism, we propose an end-to-end detail-recovery image deraining network (termed a DRD-Net) for single images. We for the first time introduce two sub-networks with a comprehensive loss function which synergize to derain and recover the lost details caused by deraining. We have three key contributions. First, we present a rain residual network to remove rain streaks from the rainy images, which combines the squeeze-and-excitation (SE) operation with residual blocks to make full advantage of spatial contextual information. Second, we design a new connection style block, named structure detail context aggregation block (SDCAB), which aggregates context feature information and has a large reception field. Third, benefiting from the SDCAB, we construct a detail repair network to encourage the lost details to return for eliminating image degradations. We have validated our approach on four recognized datasets (three synthetic and one real-world). Both quantitative and qualitative comparisons show that our approach outperforms the state-of-the-art deraining methods in terms of the deraining robustness and detail accuracy. The source code has been available for public evaluation and use on GitHub.

Figures

Figures reproduced from arXiv: 1908.10267 by the authors.

Figure 1
Figure 1. Image deraining results tested in the dataset of Rain200H. From (a)-(h): (a) the rainy image Castle, and the deraining [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. DRD-Net consists of two-sub networks, i.e., the rain removal network and the detail repair network. The first sub [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Illustration of different convolution styles. From (a)-(d): (a) direct network, (b) residual block, (c) rain residual block [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (13 more)
Figure 4
Figure 4. Figure 4: Feature maps with different weights. The images in [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Image deraining results tested in the dataset of Rain200L. From (a)-(e): (a) the input rainy image, (b) the result [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Image deraining results tested in the dataset of Rain200L. From (a)-(h): (a) the rainy image Mountain, and the [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: Image deraining results tested in the dataset of Rain200H. From (a)-(h): (a) the rainy image Zebra, and the deraining [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Image deraining results tested in the dataset of Rain200H. From (a)-(h): (a) the rainy image Match, and the deraining [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 9
Figure 9. Figure 9: Image deraining results tested in real-world dataset. From (a)-(h): (a) the rainy image Plant, and the deraining [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 10
Figure 10. Figure 10: Image deraining results tested in real-world dataset. From (a)-(h): (a) the rainy image Obama, and the deraining [PITH_FULL_IMAGE:figures/full_fig_p009_10.png]
Figure 11
Figure 11. Figure 11: Image deraining results tested in real-world dataset. From (a)-(h): (a) the rainy image Trees, and the deraining [PITH_FULL_IMAGE:figures/full_fig_p009_11.png]
Figure 12
Figure 12. Figure 12: Image deraining results tested in real-world dataset. From (a)-(h): (a) the rainy image Putin, and the deraining [PITH_FULL_IMAGE:figures/full_fig_p010_12.png]
Figure 13
Figure 13. Figure 13: Image deraining results tested in real-world dataset. From (a)-(h): (a) the rainy image House, and the deraining [PITH_FULL_IMAGE:figures/full_fig_p010_13.png]
Figure 14
Figure 14. Figure 14: Image deraining results tested in real-world dataset. From (a)-(h): (a) the rainy image Street Lamp, and the deraining [PITH_FULL_IMAGE:figures/full_fig_p011_14.png]
Figure 15
Figure 15. Figure 15: Image deraining results tested in real-world dataset. From (a)-(h): (a) the rainy image Rainforest, and the deraining [PITH_FULL_IMAGE:figures/full_fig_p011_15.png]
Figure 16
Figure 16. Figure 16: The deraining results tested on the Google Vision API. From (a)-(c): (a) object recognition result in the real-world [PITH_FULL_IMAGE:figures/full_fig_p012_16.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. SSH-Net: A Self-Supervised and Hybrid Network for Noisy Image Watermark Removal

    cs.MM 2025-05 reject novelty 4.0 of 10

    SSH-Net combines a CNN noise-removal decoder and a sparse Transformer watermark decoder with a shared encoder and gate fusion, claiming SOTA noisy watermark removal, but its self-supervised target is not an unbiased e...

Reference graph

Works this paper leans on

28 extracted references · 17 canonical work pages · cited by 1 Pith paper

  1. [1]

    Detection and removal of rain from videos,

    K. Garg and S. K. Nayar, “Detection and removal of rain from videos,” in 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR 2004), with CD-ROM, 27 June - 2 July 2004, Washington, DC, USA , 2004, pp. 528–535. [Online]. Available: http://doi.ieeecomputersociety.org/10.1109/ CVPR.2004.79 10 (a) Rainy image (h) Ours (e) RES...

    work page 2004

  2. [2]

    Utilizing local phase information to remove rain from video,

    V . Santhaseelan and V . K. Asari, “Utilizing local phase information to remove rain from video,”International Journal of Computer Vision, vol. 112, no. 1, pp. 71–89, 2015

    work page 2015

  3. [3]

    Rain streak removal using layer priors,

    Y. Li, R. T. Tan, X. Guo, J. Lu, and M. S. Brown, “Rain streak removal using layer priors,” in 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016, Las Vegas, NV , USA, June 27-30, 2016, vol. 12, 2016, pp. 2736–2744

    work page 2016

  4. [4]

    Removing rain from a single image via discriminative sparse coding,

    Y. Luo, Y. Xu, and H. Ji, “Removing rain from a single image via discriminative sparse coding,” in 2015 IEEE International Confer- ence on Computer Vision, ICCV 2015, Santiago, Chile, December 7-13, 2015, vol. 8, 2015, pp. 3397–3405

    work page 2015

  5. [5]

    Convolutional sparse and low-rank coding-based rain streak removal,

    H. Zhang and V . M. Patel, “Convolutional sparse and low-rank coding-based rain streak removal,” in 2017 IEEE Winter Conference on Applications of Computer Vision, WACV 2017, Santa Rosa, CA, USA, March 24-31, 2017 , 2017, pp. 1259–1267. [Online]. Available: https://doi.org/10.1109/WACV .2017.145

    doi:10.1109/wacv 2017

  6. [6]

    Robust low- rank subspace segmentation with finite mixture noise,

    X. Guo, X. Xie, G. Liu, M. Wei, and J. Wang, “Robust low- rank subspace segmentation with finite mixture noise,” Pattern Recognition, vol. 93, pp. 55–67, 2019

    work page 2019

  7. [7]

    Removing rain from single images via a deep detail network,

    X. Fu, J. Huang, D. Zeng, Y. Huang, X. Ding, and J. W. Paisley, “Removing rain from single images via a deep detail network,” in 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017 , 2017, pp. 1715–1723. [Online]. Available: https://doi.org/10.1109/CVPR.2017.186

    doi:10.1109/cvpr.2017.186 2017

  8. [8]

    Depth-attentional features for single-image rain removal,

    X. Hu, C.-W. Fu, L. Zhu, and P .-A. Heng, “Depth-attentional features for single-image rain removal,” in The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019

    work page 2019

Show all 28 references
  1. [10]

    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,” in Computer Vision - ECCV 2018 - 15th European Conference, Munich, Germany, September 8-14, 2018, Proceedings, Part VII , 2018, pp. 262–277. [Onlin...

    2018

  2. [11]

    Analysis of rain and snow in frequency space,

    P . C. Barnum, S. G. Narasimhan, and T. Kanade, “Analysis of rain and snow in frequency space,” International Journal of Computer Vision, vol. 86, no. 2-3, pp. 256–274, 2010. [Online]. Available: https://doi.org/10.1007/s11263-008-0200-2

    2010 doi

  3. [12]

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

    J. Bossu, N. Hauti `ere, and J. Tarel, “Rain or snow detection in image sequences through use of a histogram of orientation of streaks,” International Journal of Computer Vision , vol. 93, no. 3, pp. 348–367, 2011. [Online]. Available: https://doi.org/10.1007/ s11263-011-0421-...

    2011

  4. [13]

    Removal of rain from videos: a review,

    A. K. Tripathi and S. Mukhopadhyay, “Removal of rain from videos: a review,” Signal, Image and Video Processing , vol. 8, no. 8, pp. 1421–1430, 2014. [Online]. Available: https://doi.org/10.1007/s11760-012-0373-6

    2014 doi

  5. [14]

    Automatic single-image- based rain streaks removal via image decomposition,

    L.-W. Kang, C.-W. Lin, and Y.-H. Fu, “Automatic single-image- based rain streaks removal via image decomposition,” IEEE Trans- actions on Image Processing, vol. 21, no. 4, pp. 1742–1755, 2011

    2011

  6. [15]

    Joint bi-layer optimization for single-image rain streak removal,

    L. Zhu, C. Fu, D. Lischinski, and P . Heng, “Joint bi-layer optimization for single-image rain streak removal,” in IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, 2017, pp. 2545–2553. [Online]. Available: https://doi.org/10.1109...

    2017 doi

  7. [16]

    Deep joint rain detection and removal from a single image,

    W. Yang, R. T. Tan, J. Feng, J. Liu, Z. Guo, and S. Yan, “Deep joint rain detection and removal from a single image,” in 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017 , 2017, pp. 1685–1694. [Online]. Available: ...

    2017 doi

  8. [17]

    Image de-raining using a conditional generative adversarial network,

    H. Zhang, V . Sindagi, and V . M. Patel, “Image de-raining using a conditional generative adversarial network,” CoRR, vol. abs/1701.05957, 2017. [Online]. Available: http://arxiv.org/abs/ 1701.05957

    2017 arXiv

  9. [18]

    Joint rain detection and removal via iterative region dependent multi-task learning,

    W. Yang, R. T. Tan, J. Feng, J. Liu, Z. Guo, and S. Yan, “Joint rain detection and removal via iterative region dependent multi-task learning,” CoRR, vol. abs/1609.07769, 2016. [Online]. Available: http://arxiv.org/abs/1609.07769

    2016 arXiv

  10. [19]

    Density-aware single image de-raining using a multi-stream dense network,

    H. Zhang and V . M. Patel, “Density-aware single image de-raining using a multi-stream dense network,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2018, pp. 695–704

    2018

  11. [20]

    Residual-guide network for single image deraining,

    Z. Fan, H. Wu, X. Fu, Y. Huang, and X. Ding, “Residual-guide network for single image deraining,” in 2018 ACM Multimedia Conference on Multimedia Conference. ACM, 2018, pp. 1751–1759

    2018

  12. [21]

    Squeeze-and-excitation networks,

    J. Hu, L. Shen, and G. Sun, “Squeeze-and-excitation networks,” in 2018 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2018, Salt Lake City, UT, USA, June 18-22, 2018 , 2018, pp. 7132–7141. [Online]. Avail- able: http://openaccess.thecvf.com/content cvpr 2018/...

    2018

  13. [22]

    Batch normalization: Accelerating deep network training by reducing internal covariate shift,

    S. Ioffe and C. Szegedy, “Batch normalization: Accelerating deep network training by reducing internal covariate shift,” in Proceedings of the 32nd International Conference on Machine Learning, ICML 2015, Lille, France, 6-11 July 2015, 2015, pp. 448–456. [Online]. Available: h...

    2015

  14. [23]

    Delving deep into rectifiers: Surpassing human-level performance on imagenet classification,

    K. He, X. Zhang, S. Ren, and J. Sun, “Delving deep into rectifiers: Surpassing human-level performance on imagenet classification,” in 2015 IEEE International Conference on Computer Vision, ICCV 12 (a) (b) (c) Fig. 16: The deraining results tested on the Google Vision API. From ...

    2015

  15. [24]

    Context-aware single image rain removal,

    D. Huang, L. Kang, M. Yang, C. Lin, and Y. F. Wang, “Context-aware single image rain removal,” in Proceedings of the 2012 IEEE International Conference on Multimedia and Expo, ICME 2012, Melbourne, Australia, July 9-13, 2012 , 2012, pp. 164–169. [Online]. Available: https://do...

    2012 doi

  16. [25]

    Multi-scale context aggregation by dilated convolutions,

    F. Yu and V . Koltun, “Multi-scale context aggregation by dilated convolutions,” in 4th International Conference on Learning Representations, ICLR 2016, San Juan, Puerto Rico, May 2-4, 2016, Conference Track Proceedings , 2016. [Online]. Available: http://arxiv.org/abs/1511.07122

    2016 arXiv

  17. [26]

    Adam: A method for stochastic optimiza- tion,

    D. P . Kingma and J. Ba, “Adam: A method for stochastic optimiza- tion,” arXiv preprint arXiv:1412.6980, 2014

    2014 arXiv

  18. [27]

    Scope of validity of psnr in image/video quality assessment,

    Q. Huynh-Thu and M. Ghanbari, “Scope of validity of psnr in image/video quality assessment,” Electronics letters, vol. 44, no. 13, pp. 800–801, 2008

    2008

  19. [28]

    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,” IEEE Trans. Image Processing , vol. 13, no. 4, pp. 600–612, 2004. [Online]. Available: https://doi.org/10.1109/TIP .2003.819861

    2004

  20. [29]

    A directional global sparse model for single image rain removal,

    L.-J. Deng, T.-Z. Huang, X.-L. Zhao, and T.-X. Jiang, “A directional global sparse model for single image rain removal,” Applied Math- ematical Modelling, vol. 59, pp. 662–679, 2018. Sen Deng is now pursuing his PhD degree at the School of Computer Science and Technology, Nanj...

    2014

Pith tools

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