Pith. sign in

REVIEW 4 major objections 5 minor 62 references

Towards Context-aware Convolutional Network for Image Restoration

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

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

Pith's one-line read A 4.26M-parameter convolutional network outperforms transformer-based models on three image restoration tasks.

desk verdict A compact CNN with strong reported numbers on dehazing/deblurring/desnowing, but the load-bearing LDIM receptive-field claim is under-specified and the RSAM is vestigial; deserves review with code. read the letter →

arxiv 2412.11008 v1 pith:T4YZE22X submitted 2024-12-15 cs.CV

Fangwei Hao , Ji Du , Weiyun Liang , Jing Xu , Xiaoxuan Xu This is my paper · ORCID
classification cs.CV
keywords imagerestorationdehazingmotiondeblurringdesnowingstaroperationstripattentionreceptivefieldconvolutionalnetwork
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 argues that a purely convolutional image-restoration network can match or beat transformer-based models by combining two ideas: a residual 'star' module that multiplies two learned feature maps element-wise, and a large dynamic integration module that stacks horizontal and vertical strip attention with dilated strip attention to gather context from a very wide area. If the reported results hold, the network achieves 41.25 dB PSNR on SOTS-Indoor dehazing, 40.29 dB on SOTS-Outdoor, 33.31 dB on GoPro motion deblurring, and 38.29 dB on CSD desnowing while using roughly 4.26 million parameters. The paper's point is that long-range context, usually the selling point of transformers, can be obtained dynamically with cheap convolutional operations, making high-performance restoration feasible at low model complexity.

What carries the argument

The paper's two load-bearing mechanisms are the efficient residual star module (ERSM) and the large dynamic integration module (LDIM). The 'star operation' is element-wise multiplication of two feature vectors; ERSM makes it context-aware by feeding one branch through a large depth-wise convolution so that the multiplied features carry local contextual information before being projected into a high-dimensional nonlinear space. LDIM builds a wide square receptive field by composing horizontal strip attention (H-SA) of strip size $K$ with horizontal dilated strip attention (H-DSA) whose dilation rate is $d_r=(k+1)/2$, then repeating the pair vertically; because the first strip attention already mixes pixels along the strip, the subsequent dilated attention lets the module reach pixels far beyond a single kernel without quadratic cost. All attention weights are generated dynamically from the input by convolutions, and the module uses multi-scale receptive fields across feature groups to handle blurs of different sizes.

What would settle it

Inspect the implementation to read off $K$ and the dilation rates; if the effective receptive field of LDIM covers only a small fraction of the $256\times256$ training patches, or if removing LDIM changes PSNR by far less than the reported 2.36 dB on the SOTS-Indoor ablation, then the large-context explanation is not what carries the result.

Watch

Extended reading notes

Core claim

On its own terms, the paper claims that the efficient residual star module (ERSM) and the large dynamic integration module (LDIM), placed in a six-scale U-shaped encoder-decoder, yield a context-aware convolutional network that outperforms prior state-of-the-art restoration methods on image dehazing, motion deblurring, and desnowing. ERSM replaces a plain residual block with a context-aware star unit: one branch is a $1\times1$ convolution, the other is a $1\times1$ convolution followed by a large depth-wise convolution and GELU, and their element-wise product is refined by a $3\times3$ convolution with a skip connection. LDIM applies horizontal strip attention, then horizontal dilated strip attention, then the vertical equivalents, with all attention weights learned dynamically from the input features; the paper argues that the strip-then-dilate composition gives every operated pixel access to any input pixel over an extremely large square region. Ablations on SOTS-Indoor attribute 2.36 dB of the 2.85 dB total gain over the baseline to LDIM and show that ERSM adds another 0.49 dB, with the full model reaching 41.25 dB.

Load-bearing premise

The whole 'extremely large receptive field' argument depends on the strip size $K$ and the per-group dilation rates being large enough that the strip-then-dilated composition actually reaches across nearly the whole feature map; the paper never states these values or computes the resulting receptive field.

Editorial extensions

If this is right

  • A convolutional restoration network with roughly 4.26 million parameters can outperform transformer-based models on the dehazing and motion-deblurring benchmarks reported here, which would lower the compute barrier for practical restoration.
  • The strip-then-dilated attention composition offers a parameter-light way to obtain long-range context without quadratic self-attention, so similar modules could be dropped into other low-level vision networks.
  • Because LDIM's weights are predicted from the input and applied at multiple scales, the network can adapt its context integration to blurred regions of different sizes, matching the variability found in real motion blur.
  • The dual-domain $L_1$ loss (spatial plus frequency) combined with multi-input/multi-output supervision makes training stable enough that one U-shaped network handles three degradation types with task-specific block counts.

Reading between the lines

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

  • If the receptive-field claim is verified, the same horizontal-then-vertical dilated strip composition could be applied to other long-range tasks such as single-image deraining or super-resolution, where the degradation is also spatially varying.
  • A direct measurement of the effective receptive field, for example by gradient propagation or input perturbation, would show whether the benefit comes from true long-range integration or from the dynamic weighting alone; the paper does not provide that measurement.
  • The ablation suggests LDIM contributes most of the gain, so a natural extension would be to test LDIM alone in a simple ResNet-style backbone, without the star module, to isolate how much of the improvement is due to context integration rather than the star operation.
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

4 major / 5 minor

Summary. The paper proposes CCNet, a U-shaped convolutional network for image restoration, with two main components. The efficient residual star module (ERSM) uses element-wise multiplication ('star operation') together with a large depthwise convolution branch, and the large dynamic integration module (LDIM) combines horizontal and vertical strip attention with dilated strip attention to aggregate context. The authors report state-of-the-art PSNR/SSIM on SOTS dehazing, GoPro/HIDE motion deblurring, and CSD desnowing using about 4.3M parameters, and they provide ablations on SOTS-Indoor attributing the gains to ERSM and LDIM.

Significance. If the reported benchmark numbers are reproducible, the efficiency claim is practically significant: a 4.3M-parameter CNN outperforms several transformer-based methods, including Restormer on GoPro and DehazeFormer-L on SOTS-Indoor, while using substantially fewer parameters. The D-RSM versus ERSM comparison in Table 4 is a fair controlled test with matched parameter counts, and the multi-task evaluation across dehazing, deblurring, and desnowing is a strength. However, the paper's central architectural mechanism for LDIM is under-specified: Eq. (6) cannot be used to compute the claimed 'extremely large' receptive field, no values for K or D are given, and the proposed RSAM module in Section 3.4 is never evaluated. These issues make the significance conditional on additional clarification and verification.

major comments (4)
  1. [Section 3.3, Eq. (6)] Eq. (6) is ambiguous and prevents the reader from verifying the LDIM receptive field. The text describes multi-scale dilation applied across feature groups, but the equation sums only over k and writes d = 1,...,D without showing how d is assigned to channels or groups. As written, each output channel appears to use a single dilation rate, which does not match the group-wise description in the text and Fig. 4. The strip size K and dilation counts D are never specified anywhere in the paper, and no receptive-field size is computed. Please give the group split, report K and D, and provide an explicit formula or numerical value for the receptive field of one H-LDSI/V-LDSI and of the full LDIM.
  2. [Section 3.3, paragraphs after Eq. (8)] The claim that 'its operated pixels contain the contribution of any input pixel of the H-LDSI' and that LDIM possesses an 'extremely large square receptive field' is an overstatement. With strip size K and dilation dr = (K+1)/2, H-SA integrates over K neighboring pixels and H-DSA then samples those K positions, so each output pixel formally depends on a span of about K * dr = K(K+1)/2 pixels, not on any input pixel of the whole feature map. Unless K is large enough to cover the image, this is not global context. The paper should state the effective receptive field explicitly and temper the 'extremely large' claim accordingly, since the +2.36 dB gain of LDIM in Table 4 is attributed to this mechanism.
  3. [Section 3.4 and Table 4] The residual star attention module (RSAM) introduced in Section 3.4 is claimed to 'significantly boost model performance,' but it does not appear in the network architecture description of Section 3.1, in Fig. 2, or in any experiment or ablation in Table 4. The module is therefore unsupported by evidence. Either integrate RSAM into the actual architecture and ablate it, or remove Section 3.4 and the associated claim. As it stands, a reader cannot tell whether the reported CCNet includes RSAM at all.
  4. [Section 4.3, Table 4] The ablation evidence for the star operation's benefit is only conditional. The model with D-RSM alone (star unit without the context branch) drops to 37.86 dB, below the 38.40 dB baseline, while ERSM raises performance to 40.23 dB. This shows that the improvement comes from the added large depthwise convolutional context branch, not from the star operation per se. The abstract and Section 3.2 credit 'context-aware star operation' with high-dimensional, non-linear feature mapping; the paper should either temper this claim or add a control that isolates the star operation's contribution, for example ERSM without the star operation or with the star operation placed differently.
minor comments (5)
  1. [Table 2] The HIDE column headers read 'PSNR↓ SSIM↓' but these metrics should be the higher-the-better arrows (↑), as in the GoPro columns.
  2. [Section 4.2, Image Desnowing] The text cites the CSD dataset as [47], but reference [47] is the PyTorch software paper; the CSD dataset is defined by reference [46]. Please correct the citation.
  3. [Section 3.3, notation] The symbol k is used both for the summation index and for the strip size in dr = (k+1)/2, while the strip size is otherwise denoted K. Please use consistent notation, e.g., K for the strip size and k for the summation index.
  4. [Section 3.5, Eq. (11)] The loss equation defines s as a normalization denominator but then says 'S denotes the total elements'; the lowercase s and uppercase S are inconsistent. Use one symbol throughout.
  5. [Throughout] There are several typographical errors, including 'incluing' (Section 3.2), 'genarate' (Section 3.1), and 'ultilize' (Section 4.3). A copyedit pass is recommended.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: CCNet's headline numbers are held-out benchmark measurements; the LDIM receptive-field claim follows from the paper's own composition equations, not from a fitted quantity or self-citation.

full rationale

The paper's central claims are empirical: Tables 1-3 report PSNR/SSIM on held-out SOTS, GoPro, HIDE, and CSD test sets for networks trained on the corresponding training splits, so there is no fitted parameter subsequently renamed as a prediction. The ERSM and LDIM contributions are validated by ablations (Table 4), not derived from the performance numbers. The 'extremely large' receptive-field claim for LDIM is a compositional property of Eq. (4) followed by Eq. (6): H-SA gives each output pixel access to a K-wide strip of inputs, and H-DSA then samples those integrated features over a dilated span, so the stated claim follows from the paper's own equations rather than from circular reasoning. The high-dimensional nonlinear mapping property of the star operation is attributed to the external StarNet result [42], not to the present authors' prior work, and it functions as a design rationale rather than as a forced conclusion. The ambiguous indexing in Eq. (6), the unspecified strip size K, and the unstated per-group dilation split are genuine reproducibility/correctness concerns about whether the implemented LDIM actually has the claimed receptive field, but they do not make the derivation circular. No self-citation chain, uniqueness import, or ansatz smuggled through the authors' own prior work appears.

Assumptions & free parameters 4 free parameters · 3 assumptions · 1 invented entities

The central empirical claims rest on standard deep learning assumptions: the U-shaped backbone, the dual-domain loss, and the transferability of star operation and strip attention from prior work. The only invented entity, RSAM, is not used. The omitted hyperparameters (K, D, C) are the main gap between the paper's claims and what a reader can independently verify.

free parameters (4)
  • number of ERSM blocks N per task = 3 (dehazing), 15 (deblurring), 5 (desnowing)
    Chosen by hand per task complexity; not derived and not ablated across values.
  • LDIM strip size K and group dilation counts D = not specified
    The claimed large receptive field and multi-scale processing depend on these; their absence blocks replication and verification.
  • channel width C and LDIM group split = not specified
    Architecture capacity and the multi-scale group structure are undefined in the paper.
  • loss weight lambda = 0.1
    Set heuristically following MIMO-UNet+; no sensitivity study is provided.
assumptions (3)
  • domain assumption Star operation maps inputs into exceedingly high-dimensional, non-linear feature spaces (per StarNet, ref [42]).
    The core motivation for ERSM; borrowed from high-level vision and not re-derived for low-level restoration. The paper's own ablation shows the bare star block (D-RSM) lowers performance, so the transfer is not unconditional.
  • standard math H-SA followed by H-DSA covers every input pixel of the strip, so dilation leaves no holes.
    Stated in Sec 3.3 without proof or explicit kernel sizes; holds only for the right choice of K and d and with valid indexing in Eq (6).
  • domain assumption Dual-domain L1 loss with FFT improves restoration.
    Taken from MIMO-UNet+ [7]; no ablation isolates its contribution in CCNet.
invented entities (1)
  • RSAM (Residual Star Attention Module)
    purpose: Proposed in Sec 3.4 and Fig 5 to combine CSU, LDIM, and a 3x3 conv with residual structure.
    Defined but never used in the network architecture, experiments, or ablations; no performance evidence is given, so it is an unused design.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards Context-aware Convolutional Network for Image Restoration." pith.science (2026). https://pith.science/paper/T4YZE22X

@misc{pith2026241211008,
  author       = {Pith},
  title        = {Pith review of: Towards Context-aware Convolutional Network for Image Restoration},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/T4YZE22X}},
  note         = {Machine review of arXiv:2412.11008}
}
read the original abstract

Image restoration (IR) is a long-standing task to recover a high-quality image from its corrupted observation. Recently, transformer-based algorithms and some attention-based convolutional neural networks (CNNs) have presented promising results on several IR tasks. However, existing convolutional residual building modules for IR encounter limited ability to map inputs into high-dimensional and non-linear feature spaces, and their local receptive fields have difficulty in capturing long-range context information like Transformer. Besides, CNN-based attention modules for IR either face static abundant parameters or have limited receptive fields. To address the first issue, we propose an efficient residual star module (ERSM) that includes context-aware "star operation" (element-wise multiplication) to contextually map features into exceedingly high-dimensional and non-linear feature spaces, which greatly enhances representation learning. To further boost the extraction of contextual information, as for the second issue, we propose a large dynamic integration module (LDIM) which possesses an extremely large receptive field. Thus, LDIM can dynamically and efficiently integrate more contextual information that helps to further significantly improve the reconstruction performance. Integrating ERSM and LDIM into an U-shaped backbone, we propose a context-aware convolutional network (CCNet) with powerful learning ability for contextual high-dimensional mapping and abundant contextual information. Extensive experiments show that our CCNet with low model complexity achieves superior performance compared to other state-of-the-art IR methods on several IR tasks, including image dehazing, image motion deblurring, and image desnowing.

Figures

Figures reproduced from arXiv: 2412.11008 by the authors.

Figure 1
Figure 1. Performance and parameters of different methods on SOTS-Outdoor dataset. complexity of self-attention with respect to spatial resolution of images. In order to alleviate the complexity, researchers have proposed some certain algorithms to improve the IR efficiency of transformer-based methods. Concretely, both SwinIR [13] and Uformer [14] simplify the IR Transformer models by re￾stricting the spatial zone of self-at… view at source ↗
Figure 2
Figure 2. Network architecture of our CCNet. [9], have been introduced or developed from high-level tasks. Furthermore, Transformer-based models [12], [13], [14], [15], [22] have also been employed on low-level tasks to better cap￾ture long-range interdependence, and they have substantially improved the state-of-the-art performance of IR tasks. Specifi￾cally, Guo et al. [22] first propose a DeHamer which introduces Transforme… view at source ↗
Figure 3
Figure 3. The details of different modules. (a) Lightweight star block (StarB) in [42] for high-level tasks. (b) The proposed efficient residual star module (ERSM) which includes context-aware star operation for low-level tasks. (c) The designed D-RSM for comparison. focus on learning the residual information. In addition, similar to previous methods [7], [19], [35], we also adopt multi-input and multi-output strategies to ea… view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: The details of the proposed horizontal large dynamic strip integration (H-LDSI), vertical large dynamic strip integration (V-LDSI), and large dynamic [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: The details of the proposed context-aware residual star attention mod [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Comparisons of Image dehazing on the SOTS-Indoor. The best result is highlighted. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Comparisons of Image dehazing on the SOTS-Outdoor. The best result is highlighted. [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Comparisons of image motion deblurring on the GoPro dataset. The best result is highlighted. [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Comparisons of Image desnowing on the CSD dataset. The best results are [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]
Figure 10
Figure 10. Figure 10: Performance and the parameters of different methods on SOTS￾Indoor dataset. 4.4. Model Complexity Analysis In order to further validate the effectiveness and efficiency of our CCNet, we analyze the model complexity and make a com￾parison with other previous state-of-t…
Figure 11
Figure 11. Figure 11: Performance and the Multi-Adds of different methods on SOTS￾Indoor dataset. ful learning ability for contextual high-dimensional mapping and abundant contextual information integration. Specifically, we propose an efficient residual star module (ERSM) which includes c…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

62 extracted references · 57 canonical work pages

  1. [1]

    W., Arora, A., Khan, S., Hayat, M., Khan, F

    Zamir, S. W., Arora, A., Khan, S., Hayat, M., Khan, F. S., & Yang, M. H. (2022). Restormer: E fficient transformer for high-resolution im- age restoration. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 5728-5739)

    work page 2022

  2. [2]

    Su, J., Xu, B., & Yin, H. (2022). A survey of deep learning approaches to image restoration. Neurocomputing, 487, 46-65

    work page 2022

  3. [3]

    S., Stenger, B., Yang, M

    Zhang, K., Ren, W., Luo, W., Lai, W. S., Stenger, B., Yang, M. H., & Li, H. (2022). Deep image deblurring: A survey. International Journal of Computer Vision, 130(9), 2103-2130

    work page 2022

  4. [4]

    Hunt. (1977). Bayesian methods in nonlinear digital image restoration. IEEE Transactions on Computers, 100(3), 219-229

    work page 1977

  5. [5]

    L., & Eom, K

    Kashyap, R. L., & Eom, K. B. (1988). Robust image modeling techniques with an image restoration application. IEEE Transactions on Acoustics, Speech, and Signal Processing, 36(8), 1313-1325

    work page 1988

  6. [6]

    A., & Nowak, R

    Figueiredo, M. A., & Nowak, R. D. (2003). An EM algorithm for wavelet- based image restoration. IEEE Transactions on Image Processing,12(8), 906-916

    work page 2003

  7. [8]

    Lee, J., Son, H., Rim, J., Cho, S., & Lee, S. (2021). Iterative filter adap- tive network for single image defocus deblurring. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 2034-2042)

    work page 2021

  8. [9]

    Zou, W., Jiang, M., Zhang, Y ., Chen, L., Lu, Z., & Wu, Y . (2021). Sd- wnet: A straight dilated network with wavelet transformation for image deblurring. In Proceedings of the IEEE/CVF international conference on computer vision (pp. 1895-1904)

    work page 2021

Show all 62 references
  1. [11]

    Cui, Y ., Ren, W., Yang, S., Cao, X., & Knoll, A. (2023). Irnext: Rethink- ing convolutional network design for image restoration. In International conference on machine learning

    2023

  2. [12]

    & Gao, W

    Chen, H., Wang, Y ., Guo, T., Xu, C., Deng, Y ., Liu, Z., ... & Gao, W. (2021). Pre-trained image processing transformer. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 12299-12310)

    2021

  3. [13]

    Liang, J., Cao, J., Sun, G., Zhang, K., Van Gool, L., & Timofte, R. (2021). Swinir: Image restoration using swin transformer. In Proceedings of the IEEE/CVF international conference on computer vision (pp. 1833-1844)

    2021

  4. [14]

    Wang, Z., Cun, X., Bao, J., Zhou, W., Liu, J., & Li, H. (2022). Uformer: A 10 general u-shaped transformer for image restoration. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 17683-17693)

    2022

  5. [15]

    J., Peng, Y

    Tsai, F. J., Peng, Y . T., Lin, Y . Y ., Tsai, C. C., & Lin, C. W. (2022, Oc- tober). Stripformer: Strip transformer for fast image deblurring. In Eu- ropean Conference on Computer Vision (pp. 146-162). Cham: Springer Nature Switzerland

    2022

  6. [16]

    (2023, July)

    Luo, P., Xiao, G., Gao, X., & Wu, S. (2023, July). LKD-Net: Large ker- nel convolution network for single image dehazing. In 2023 IEEE Inter- national Conference on Multimedia and Expo (ICME) (pp. 1601-1606). IEEE

    2023

  7. [17]

    P., Myszkowski, K., & Chen, B

    Ruan, L., Bemana, M., Seidel, H. P., Myszkowski, K., & Chen, B. (2023). Revisiting image deblurring with an e fficient ConvNet. arXiv preprint arXiv:2302.02234

    2023 arXiv

  8. [18]

    Wang, Y ., Li, Y ., Wang, G., & Liu, X. (2024). Multi-scale attention net- work for single image super-resolution. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 5950- 5960)

    2024

  9. [19]

    Cui, Y ., Tao, Y ., Jing, L., & Knoll, A. (2023). Strip attention for image restoration. In International Joint Conference on Artificial Intelligence, IJCAI

    2023

  10. [20]

    (2024, March)

    Cui, Y ., Ren, W., & Knoll, A. (2024, March). Omni-Kernel Network for Image Restoration. In Proceedings of the AAAI Conference on Artificial Intelligence (V ol. 38, No. 2, pp. 1426-1434)

    2024

  11. [21]

    Li, B., Ren, W., Fu, D., Tao, D., Feng, D., Zeng, W., & Wang, Z. (2018). Benchmarking single-image dehazing and beyond. IEEE Transactions on Image Processing, 28(1), 492-505

    2018

  12. [22]

    L., Yan, Q., Anwar, S., Cong, R., Ren, W., & Li, C

    Guo, C. L., Yan, Q., Anwar, S., Cong, R., Ren, W., & Li, C. (2022). Image dehazing transformer with transmission-aware 3d position embedding. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 5812-5820)

    2022

  13. [23]

    Tu, Z., Talebi, H., Zhang, H., Yang, F., Milanfar, P., Bovik, A., & Li, Y . (2022). Maxim: Multi-axis mlp for image processing. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 5769-5780)

    2022

  14. [24]

    (2022, October)

    Ye, T., Zhang, Y ., Jiang, M., Chen, L., Liu, Y ., Chen, S., & Chen, E. (2022, October). Perceiving and modeling density for image dehazing. In European conference on computer vision (pp. 130-145). Cham: Springer Nature Switzerland

    2022

  15. [25]

    Ren, W., Liu, S., Zhang, H., Pan, J., Cao, X., & Yang, M. H. (2016). Single image dehazing via multi-scale convolutional neural networks. In Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part II 14 (pp. 154- ...

    2016

  16. [26]

    Ren, W., Ma, L., Zhang, J., Pan, J., Cao, X., Liu, W., & Yang, M. H. (2018). Gated fusion network for single image dehazing. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 3253-3261)

    2018

  17. [27]

    & Knoll, A

    Cui, Y ., Tao, Y ., Bing, Z., Ren, W., Gao, X., Cao, X., ... & Knoll, A. (2022, September). Selective frequency network for image restoration. In The Eleventh International Conference on Learning Representations

    2022

  18. [28]

    Ronneberger, O., Fischer, P., & Brox, T. (2015). U-net: Convolutional networks for biomedical image segmentation. In Medical image comput- ing and computer-assisted intervention–MICCAI 2015: 18th international conference, Munich, Germany, October 5-9, 2015, proceedings, part I...

    2015

  19. [29]

    (2023, June)

    Chen, S., Ye, T., Liu, Y ., Liao, T., Jiang, J., Chen, E., & Chen, P. (2023, June). Msp-former: Multi-scale projection transformer for single image desnowing. In ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 1-5). IEEE

    2023

  20. [31]

    Liu, X., Ma, Y ., Shi, Z., & Chen, J. (2019). Griddehazenet: Attention- based multi-scale network for image dehazing. In Proceedings of the IEEE/CVF international conference on computer vision (pp. 7314-7323)

    2019

  21. [32]

    (2020, April)

    Qin, X., Wang, Z., Bai, Y ., Xie, X., & Jia, H. (2020, April). FFA-Net: Feature fusion attention network for single image dehazing. In Proceed- ings of the AAAI conference on artificial intelligence (V ol. 34, No. 07, pp. 11908-11915)

    2020

  22. [33]

    W., Arora, A., Khan, S., Hayat, M., Khan, F

    Zamir, S. W., Arora, A., Khan, S., Hayat, M., Khan, F. S., Yang, M. H., & Shao, L. (2021). Multi-stage progressive image restoration. In Pro- ceedings of the IEEE /CVF conference on computer vision and pattern recognition (pp. 14821-14831)

    2021

  23. [34]

    & Guo, B

    Liu, Z., Lin, Y ., Cao, Y ., Hu, H., Wei, Y ., Zhang, Z., ... & Guo, B. (2021). Swin transformer: Hierarchical vision transformer using shifted windows. In Proceedings of the IEEE /CVF international conference on computer vision (pp. 10012-10022)

    2021

  24. [35]

    Mao, X., Liu, Y ., Shen, W., Li, Q., & Wang, Y . (2021). Deep resid- ual fourier transformation for single image deblurring. arXiv preprint arXiv:2111.11745, 2(3), 5

    2021 arXiv

  25. [36]

    S., Thomas, A.,

    Fu, D., Arora, S., Grogan, J., Johnson, I., Eyuboglu, E. S., Thomas, A., ... & R´e, C. (2024). Monarch mixer: A simple sub-quadratic gemm-based architecture. Advances in Neural Information Processing Systems, 36

    2024

  26. [37]

    Y ., Dao, T., Baccus, S.,

    Poli, M., Massaroli, S., Nguyen, E., Fu, D. Y ., Dao, T., Baccus, S., ... & R ´e, C. (2023, July). Hyena hierarchy: Towards larger convolutional language models. In International Conference on Machine Learning (pp. 28043-28078). PMLR

    2023

  27. [38]

    Gu, A., & Dao, T. (2023). Mamba: Linear-time sequence modeling with selective state spaces. arXiv preprint arXiv:2312.00752

    2023 arXiv

  28. [39]

    N., & Lu, J

    Rao, Y ., Zhao, W., Tang, Y ., Zhou, J., Lim, S. N., & Lu, J. (2022). Hornet: Efficient high-order spatial interactions with recursive gated convolutions. Advances in Neural Information Processing Systems, 35, 10353-10366

    2022

  29. [40]

    H., Lu, C

    Guo, M. H., Lu, C. Z., Liu, Z. N., Cheng, M. M., & Hu, S. M. (2023). Visual attention network. Computational Visual Media, 9(4), 733-752

    2023

  30. [41]

    Yang, J., Li, C., Dai, X., & Gao, J. (2022). Focal modulation networks. Advances in Neural Information Processing Systems, 35, 4203-4217

    2022

  31. [42]

    Ma, X., Dai, X., Bai, Y ., Wang, Y ., & Fu, Y . (2024). Rewrite the Stars. In Proceedings of the IEEE /CVF Conference on Computer Vision and Pattern Recognition (pp. 5694-5703)

    2024

  32. [43]

    P., & Ba, J

    Kingma, D. P., & Ba, J. (2014). Adam: A method for stochastic optimiza- tion. arXiv preprint arXiv:1412.6980

    2014 arXiv

  33. [44]

    Nah, S., Hyun Kim, T., & Mu Lee, K. (2017). Deep multi-scale con- volutional neural network for dynamic scene deblurring. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 3883-3891)

    2017

  34. [45]

    Shen, Z., Wang, W., Lu, X., Shen, J., Ling, H., Xu, T., & Shao, L. (2019). Human-aware motion deblurring. In Proceedings of the IEEE/CVF Inter- national Conference on Computer Vision (pp. 5572-5581)

    2019

  35. [46]

    T., Fang, H

    Chen, W. T., Fang, H. Y ., Hsieh, C. L., Tsai, C. C., Chen, I., Ding, J. J., & Kuo, S. Y . (2021). All snow removed: Single image desnowing al- gorithm using hierarchical dual-tree complex wavelet representation and contradict channel loss. In Proceedings of the IEEE /CVF Inte...

    2021

  36. [47]

    & Lerer, A

    Paszke, A., Gross, S., Chintala, S., Chanan, G., Yang, E., DeVito, Z., ... & Lerer, A. (2017). Automatic differentiation in pytorch

    2017

  37. [48]

    He, K., Sun, J., & Tang, X. (2010). Single image haze removal using dark channel prior. IEEE transactions on pattern analysis and machine intelligence, 33(12), 2341-2353

    2010

  38. [49]

    & Hua, G

    Chen, D., He, M., Fan, Q., Liao, J., Zhang, L., Hou, D., ... & Hua, G. (2019, January). Gated context aggregation network for image dehazing and deraining. In 2019 IEEE winter conference on applications of com- puter vision (W ACV) (pp. 1375-1383). IEEE

    2019

  39. [50]

    Dong, H., Pan, J., Xiang, L., Hu, Z., Zhang, X., Wang, F., & Yang, M. H. (2020). Multi-scale boosted dehazing network with dense feature fu- sion. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 2157-2167)

    2020

  40. [51]

    Dong, J., & Pan, J. (2020). Physics-based feature dehazing networks. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XXX 16 (pp. 188-204). Springer International Publishing

    2020

  41. [52]

    Wu, H., Qu, Y ., Lin, S., Zhou, J., Qiao, R., Zhang, Z., ... & Ma, L. (2021). Contrastive learning for compact single image dehazing. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 10551-10560)

    2021

  42. [53]

    Song, Y ., He, Z., Qian, H., & Du, X. (2023). Vision transformers for sin- gle image dehazing. IEEE Transactions on Image Processing, 32, 1927- 1941

    2023

  43. [54]

    Kupyn, O., Martyniuk, T., Wu, J., & Wang, Z. (2019). Deblurgan-v2: Deblurring (orders-of-magnitude) faster and better. In Proceedings of the IEEE/CVF international conference on computer vision (pp. 8878-8887). 11

    2019

  44. [55]

    Zhang, K., Luo, W., Zhong, Y ., Ma, L., Stenger, B., Liu, W., & Li, H. (2020). Deblurring by realistic blurring. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 2737-2746)

    2020

  45. [56]

    Zhang, H., Dai, Y ., Li, H., & Koniusz, P. (2019). Deep stacked hierar- chical multi-patch network for image deblurring. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 5978-5986)

    2019

  46. [57]

    N., & Boddeti, V

    Purohit, K., Suin, M., Rajagopalan, A. N., & Boddeti, V . N. (2021). Spatially-adaptive image restoration using distortion-guided networks. In Proceedings of the IEEE /CVF international conference on computer vi- sion (pp. 2309-2319)

    2021

  47. [58]

    J., Ji, S

    Cho, S. J., Ji, S. W., Hong, J. P., Jung, S. W., & Ko, S. J. (2021). Rethink- ing coarse-to-fine approach in single image deblurring. In Proceedings of the IEEE/CVF international conference on computer vision (pp. 4641- 4650)

    2021

  48. [59]

    Chen, L., Lu, X., Zhang, J., Chu, X., & Chen, C. (2021). Hinet: Half instance normalization network for image restoration. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 182-192)

    2021

  49. [60]

    F., Jaw, D

    Liu, Y . F., Jaw, D. W., Huang, S. C., & Hwang, J. N. (2018). Desnownet: Context-aware deep network for snow removal. IEEE Transactions on Im- age Processing, 27(6), 3064-3073

    2018

  50. [61]

    Engin, D., Genc ¸, A., & Kemal Ekenel, H. (2018). Cycle-dehaze: En- hanced cyclegan for single image dehazing. In Proceedings of the IEEE conference on computer vision and pattern recognition workshops (pp. 825-833)

    2018

  51. [62]

    T., & Cheong, L

    Li, R., Tan, R. T., & Cheong, L. F. (2020). All in one bad weather removal using architectural search. In Proceedings of the IEEE /CVF conference on computer vision and pattern recognition (pp. 3175-3185)

    2020

  52. [63]

    T., Fang, H

    Chen, W. T., Fang, H. Y ., Ding, J. J., Tsai, C. C., & Kuo, S. Y . (2020). JS- TASR: Joint size and transparency-aware snow removal algorithm based on modified partial convolution and veiling e ffect removal. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK,...

    2020

  53. [64]

    Valanarasu, J. M. J., Yasarla, R., & Patel, V . M. (2022). Transweather: Transformer-based restoration of images degraded by adverse weather conditions. In Proceedings of the IEEE /CVF Conference on Computer Vision and Pattern Recognition (pp. 2353-2363)

    2022

  54. [65]

    (2022, October)

    Chen, L., Chu, X., Zhang, X., & Sun, J. (2022, October). Simple baselines for image restoration. In European conference on computer vision (pp. 17-33). Cham: Springer Nature Switzerland. 12

    2022

Pith tools

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