Pith. sign in

REVIEW 6 major objections 6 minor 55 references

SAAT: Synergistic Alternating Aggregation Transformer for Image Super-Resolution

T0 review · 6 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read SAAT claims a super-resolution Transformer that beats SwinIR and HAT at equal parameter count by alternating channel and spatial attention with shifted windows.

desk verdict A reasonable architecture combination undermined by inconsistent baseline numbers and a table that contradicts its own text; not ready for publication. read the letter →

arxiv 2506.03740 v1 pith:C7R44HXD submitted 2025-06-04 cs.CV cs.AI

Jianfeng Wu , Nannan Xu This is my paper
classification cs.CVcs.AI
keywords imagesuper-resolutionTransformerefficientchannelattentionmulti-scalespatialshiftedwindowoverlappingcross-attentionsinglePSNRandSSIM
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper proposes SAAT, a Transformer for single-image super-resolution whose deep feature extractor alternates between two attention groups. The CWSAG group pairs shifted-window self-attention with a cheap efficient channel attention block, and the SWSAG group pairs it with a shareable multi-scale spatial attention block; an overlapping cross-attention block closes each group. Its central claim is that this alternation recovers global channel information and spatial structural detail that window-only attention discards, and that it pushes PSNR and SSIM above the compared models on nearly every scale of Set5, Set14, BSD100, Urban100, and Manga109 at the same parameter count. A sympathetic reader would care because it suggests the synergy between channel and spatial attention, rather than simply more depth or parameters, is a practical route to sharper texture and edge reconstruction.

What carries the argument

The load-bearing mechanism is the alternating pair of attention groups. CWSAG computes (S)W-MSA in parallel with ECAB, whose average-pooled features are reweighted by a fast 1D convolution with an adaptively chosen kernel size, adding global channel context without a dimension-reducing bottleneck. SWSAG computes (S)W-MSA in parallel with SMSAB, which averages features along height and width, splits them into four parts, applies shared depthwise 1D convolutions with kernels 3, 5, 7, and 9, concatenates the parts through GroupNorm, and multiplies the resulting horizontal and vertical attention masks onto the features. Each group ends with an Overlapping Cross-Attention Block so adjacent windows exchange information, and the whole stack alternates SWSAG and CWSAG three times. The scalar factors $\alpha$ and $\beta$, both set to 0.01 in the experiments, temper the parallel branches so they do not fight the window-attention branch during optimization.

What would settle it

Evaluate the released SwinIR, HAT, SRFormer, and EDT checkpoints with the exact evaluation code used for SAAT on Set5, Set14, BSD100, Urban100, and Manga109, and retrain SAAT in the same training loop; if SAAT's reported margin over HAT and SwinIR at equal parameters drops to within run-to-run noise, roughly 0.05 dB, the paper's central claim fails.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that standard window self-attention wastes useful signal, and that two parallel attention branches placed in alternating groups can recover it without a parameter blow-up. In the Efficient Channel & Window Synergistic Attention Group, (S)W-MSA is computed in parallel with ECAB, whose average-pooled features pass through a fast 1D convolution with an adaptive kernel size to reweight channels. In the Spatial & Window Synergistic Attention Group, (S)W-MSA is computed in parallel with SMSAB, which averages the input along height and width, splits each into four parts, runs shared depthwise 1D convolutions with kernels 3, 5, 7, and 9, and multiplies the resulting vertical and horizontal attention maps onto the features. The paper reports the best numbers in its table on almost all scales, for example 28.12 dB on Urban100 at scale 4 versus 27.46 dB for HAT, and interprets the margin as evidence that alternating channel and spatial attention with window attention is what drives accuracy.

Load-bearing premise

The claim that SAAT beats the compared models at equal parameters rests on the assumption that every baseline in Table 1 was trained and evaluated under the same protocol as SAAT, including the same data, patch size, augmentation, iterations, and metric code, which the paper does not demonstrate and which its SRFormer entries contradict.

Editorial extensions

If this is right

  • At the same parameter budget, SAAT reports PSNR gains over SwinIR of roughly 0.3 to 0.7 dB across the tested scales, with the largest improvements on the texture-heavy Urban100 and Manga109 sets.
  • Because the channel branch uses an adaptive 1D convolution rather than a channel-reduction bottleneck, the reported gains come without adding a large parameter overhead compared with standard channel attention.
  • The reported visual comparisons show cleaner edges and repeated textures, indicating that the multi-scale spatial branch contributes structural detail rather than only nudging a metric.
  • If reproduced, the alternating-group design offers a way to improve transformer-based restoration that does not depend on the same-task pretraining strategy used by HAT.

Reading between the lines

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

  • If the reported gains hold under a controlled re-run, the alternation implies that channel and spatial attention need not be merged inside a single block; sequencing them in separate groups may be enough, a claim the paper does not directly test.
  • The same alternating-group recipe is a plausible candidate for other image restoration tasks such as denoising and JPEG artifact removal, which the paper does not experiment with.
  • The Table 1 numbers for SRFormer at scale 3 are far below that model's published results, so re-evaluating with official checkpoints is the quickest check on whether the state-of-the-art comparison is reliable.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

6 major / 6 minor

Summary. The paper proposes SAAT, a Transformer-based single-image super-resolution model that alternates between groups combining shifted-window multi-head self-attention with efficient channel attention (CWSAG) and with a multi-scale shared spatial attention block (SWSAG), adding overlapping cross-attention blocks and a ConvFFN. The authors claim that SAAT achieves performance comparable to state-of-the-art (SOTA) under the same parameter count, and they support this with quantitative and visual comparisons on Set5, Set14, BSD100, Urban100, and Manga109 in Table 1, plus ablation studies on Urban100. The central evidence is Table 1, which reports SAAT as best on most datasets and scales.

Significance. If the reported results were reliable, the architecture would represent a plausible incremental combination of window attention, channel attention, and spatial attention for SR, and the claimed parity with, or improvement over, SwinIR and HAT at equal parameter budgets would be of interest to the low-level vision community. However, the empirical foundation is not trustworthy: the SRFormer baseline in Table 1 is inconsistent with published values at ×3 and ×4 while matching at ×2; the text and the table contradict each other on the size of the SwinIR gain; the ablation table does not reconcile with the main table; and hyperparameters are tuned on a test set. The paper also reports no parameter counts for any model and releases no code or checkpoints, so the central 'same quantity of parameters' claim is unverifiable. The core architectural idea is plausible but not new in isolation, and the paper does not convincingly demonstrate its contribution.

major comments (6)
  1. [Table 1, SRFormer rows] The SRFormer baseline is internally inconsistent with published results. At ×2, the reported values (e.g., Set5 38.51, Urban100 34.09, Manga109 40.07) match the published SRFormer numbers exactly, but at ×3 and ×4 they fall roughly 0.5–1.5 dB below the published values (e.g., ×4 Set5 32.19 vs 33.00, Urban100 26.39 vs 27.69, Manga109 30.75 vs 32.24), even falling below the published light variant on every ×4 dataset. Since the SwinIR, HAT, and EDT rows in the same table match their published numbers, this is not a uniform protocol difference. This specifically depresses a key transformer baseline and inflates SAAT's apparent margins, so the SOTA comparison and the abstract's central claim are not supported.
  2. [§4.4.1 vs Table 1] The text states that SAAT improves over SwinIR by 0.10 dB on Urban100 and Manga109, but Table 1 shows differences of 0.69 dB (×2 Urban100: 34.50 vs 33.81), 0.67 dB (×4 Urban100: 28.12 vs 27.45), and 0.51–0.54 dB on Manga109 at ×2/×4. These two claims cannot both be correct, indicating an error in either the text or the table.
  3. [Table 2 vs Table 1] The ablation in Table 2 reports that the full model with both SMSAB and ECAB achieves PSNR/SSIM 28.47/0.8459 on Urban100, yet Table 1 lists SAAT at ×4 Urban100 as 28.12/0.8391. If both evaluations use the same Urban100 test set and the same protocol, the main result is lower than the ablation's 'best' configuration, which is a direct inconsistency that needs explanation.
  4. [§4.3.2–§4.3.3, Tables 3–4] The hyperparameters α, β, and μ are selected by maximizing PSNR on Urban100 (Tables 3 and 4), and Urban100 is also one of the evaluation datasets in Table 1. Part of the reported Urban100 gain is therefore a tuned result rather than a predicted performance, so the Urban100 margins in Table 1 are not an honest measure of generalization.
  5. [Abstract and §4.2] The central claim, 'performance comparable to the state-of-the-art under the same quantity of parameters,' is unverifiable because no parameter counts are reported for SAAT or any baseline. Given the architecture description (24 blocks, each containing (S)W-MSA, OCAB, an MLP, and an ECA or SMSAB branch), it is not obvious that the parameter count is comparable to HAT or SwinIR; the authors need to provide exact parameter counts and FLOPs to substantiate the claim.
  6. [§4.2 and Table 1] No training or evaluation details are given for any baseline method, nor is any code or checkpoint released. The paper does not state which official implementations, patch sizes, augmentation schemes, or evaluation codes were used for EDSR, RCAN, SwinIR, SRFormer, EDT, or HAT. Given the SRFormer discrepancy, the fairness of the comparison cannot be assessed, and the reader cannot reproduce the baseline numbers.
minor comments (6)
  1. [Eqs. (3)–(4)] The notation reuses f_LN for both the first and second LayerNorm without distinguishing them; this makes the equations ambiguous and should be fixed with subscripts such as f_LN^1 and f_LN^2.
  2. [Eqs. (5)–(6)] The left-hand side and the right-hand side both use the symbol X_i^H (or X_i^W) with different meanings; different symbols should be used for the input sub-feature and the convolved output.
  3. [§3.3.1, SMSAB description] The text says 'we first average the input X across each dimension to obtain two one-dimensional sequences: X_H ∈ ℝ^{B×C×H×W} and X_W ∈ ℝ^{B×C×H×W}.' Averaging should produce lower-dimensional tensors (e.g., B×C×H and B×C×W); the stated shapes are inconsistent with the description.
  4. [Throughout] Several typographical errors appear: 'Flicker2K' should be 'Flickr2K' (§4.1), 'SAA T' and 'HA T' are artifacts in Table 1, and 'LayNorm' appears in Figure 1.
  5. [Related work and §3] Reference [40] (SCSA) is cited for the synergistic use of spatial and channel attention but is not discussed in the related-work section; the relationship between SCSA and the proposed alternating SWSAG/CWSAG design should be clarified to position the novelty.
  6. [§5 Conclusion] The conclusion says SAAT is 'outperforming state-of-the-art techniques,' while the abstract claims only 'comparable to the state-of-the-art'; these claims should be reconciled.

Circularity Check

1 steps flagged · score 4.0 of 10

Partial evaluation circularity: α, β, and μ are tuned on Urban100, and the same Urban100 results are then reported as SOTA evidence in Table 1; the derivation itself is otherwise self-contained.

  1. fitted input called prediction [Section 4.3.1-4.3.3 (ablations) together with Section 4.1 and Table 1]
    "Experiments were conducted on the Urban100 dataset to evaluate PSNR/SSIM. ... we measured the model outcomes for four different values of α and β (0, 1, 0.1, 0.01). ... these experiments determined the optimal values for α and β to be 0.01 and 0.01, respectively. ... The results shown in the table 4 indicate that the model provides optimal performance at μ=0.5."

    The final SOTA table (Table 1) includes Urban100, and the model's hyperparameters α, β, and μ were chosen by grid search maximizing PSNR on that same dataset (Tables 3-4). The reported Urban100 numbers are therefore the selected optimum of the tuning objective, not an independent prediction of a held-out benchmark. This makes the Urban100 portion of the 'best performance on almost all scales' claim partially forced by construction. The other four benchmarks (Set5, Set14, BSD100, Manga109) were not used for this selection, so the central architecture claim retains independent content outside Urban100.

full rationale

No derivation-level circularity is present: the architecture equations (Eqs. 2-15) define attention blocks from stated components, and the improvement claim is empirical rather than obtained by substituting the conclusion into the premises. There are no load-bearing self-citations; the references to SwinIR, HAT, ECA-Net, and SCSA are external prior work. The one genuine circular element is evaluation-side: hyperparameters α, β, and μ were selected on Urban100 and Urban100 is then reported in Table 1 as evidence of SOTA performance. That does not invalidate the other dataset rows, so the overall circularity is partial (score 4). Two additional verification risks noted in the manuscript are correctness issues rather than circularity: Table 1 does not report parameter counts despite the 'same quantity of parameters' qualifier, and the SRFormer rows at ×3/×4 fall well below published values, which would undermine the baseline comparison if unrepaired.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The paper uses no derivations; it relies on empirical benchmarking. The central claim depends on free architectural hyperparameters, several of which were tuned on a test set, and on unverified assumptions about comparable training and evaluation of baselines.

free parameters (5)
  • alpha = 0.01
    Weight for SMSAB in SWSAB (Eq. 3), tuned by ablation in Table 3 on Urban100.
  • beta = 0.01
    Weight for ECAB in CWSAB (Eq. 11), tuned by ablation in Table 3 on Urban100.
  • mu = 0.5
    Overlap ratio in OCAB (Eq. 10), tuned by ablation in Table 4 on Urban100.
  • gamma and b = 2 and 1
    Parameters in the ECA adaptive kernel size formula (Eq. 15), taken from ECA-Net and fixed; they control the local cross-channel interaction range.
  • window size G, channels C, heads = 16, 180, 6
    Core architectural hyperparameters chosen without a reported search or justification.
assumptions (3)
  • domain assumption PSNR and SSIM on the five standard benchmarks are valid proxies for super-resolution quality.
    The paper's central comparison relies on these metrics without questioning their perceptual validity.
  • standard math ECA-Net's adaptive kernel formula (Eq. 15) transfers correctly to super-resolution features of dimension 180.
    Borrowed from ECA-Net; the authors assume the formula works in their setting without further validation.
  • domain assumption The training protocol for all baselines listed in Table 1 is comparable to that for SAAT.
    Implied by Table 1, but no baseline training details are given; the SRFormer numbers suggest this assumption may be violated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SAAT: Synergistic Alternating Aggregation Transformer for Image Super-Resolution." pith.science (2026). https://pith.science/paper/C7R44HXD

@misc{pith2026250603740,
  author       = {Pith},
  title        = {Pith review of: SAAT: Synergistic Alternating Aggregation Transformer for Image Super-Resolution},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/C7R44HXD}},
  note         = {Machine review of arXiv:2506.03740}
}
read the original abstract

Single image super-resolution is a well-known downstream task which aims to restore low-resolution images into high-resolution images. At present, models based on Transformers have shone brightly in the field of super-resolution due to their ability to capture long-term dependencies in information. However, current methods typically compute self-attention in nonoverlapping windows to save computational costs, and the standard self-attention computation only focuses on its results, thereby neglecting the useful information across channels and the rich spatial structural information generated in the intermediate process. Channel attention and spatial attention have, respectively, brought significant improvements to various downstream visual tasks in terms of extracting feature dependency and spatial structure relationships, but the synergistic relationship between channel and spatial attention has not been fully explored yet.To address these issues, we propose a novel model. Synergistic Alternating Aggregation Transformer (SAAT), which can better utilize the potential information of features. In SAAT, we introduce the Efficient Channel & Window Synergistic Attention Group (CWSAG) and the Spatial & Window Synergistic Attention Group (SWSAG). On the one hand, CWSAG combines efficient channel attention with shifted window attention, enhancing non-local feature fusion, and producing more visually appealing results. On the other hand, SWSAG leverages spatial attention to capture rich structured feature information, thereby enabling SAAT to more effectively extract structural features.Extensive experimental results and ablation studies demonstrate the effectiveness of SAAT in the field of super-resolution. SAAT achieves performance comparable to that of the state-of-the-art (SOTA) under the same quantity of parameters.

Figures

Figures reproduced from arXiv: 2506.03740 by the authors.

Figure 1
Figure 1. The overall architecture and all internal units of SAAT. based on Transformer networks. We propose Spatial & Win￾dow Synergistic Attention (SWSA) and Efficient Channel & Window Synergistic Attention (CWSA). When coordinated with window-based attention, our method integrates local and global information well and achieves good results in the task of image super-resolution. 3. Method 3.1. Overall Architecture As shown … view at source ↗
Figure 2
Figure 2. Shareable Multi-scale Spatial Attention Block(SMSAB). structure that combines SWSAG and CWSAG to enhance their interaction. As shown in [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Visual comparison on ×4 SR. PSNR/SSIM is calculated in patches marked with red boxes in the images. cross-attention. We adjust the 𝜇 value from 0 to 0.75 to study the impact of various overlap ratios, and the results are shown in the table. Here, 𝜇 = 0 corresponds to a standard Transformer block. The results shown in the table 4 indicate that the model provides optimal performance at 𝜇 = 0.5 [PITH_FULL_IMAGE:figure… view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

55 extracted references · 50 canonical work pages

  1. [1]

    Deep learning for image super-resolution: A survey

    Wang, Zhihao, Chen, Jian, Hoi, Steven C.H., 2020a. Deep learning for image super-resolution: A survey. IEEE Trans. Pattern Anal. Mach. Intell. 43 (10), 3365–3387

    work page

  2. [2]

    A deep journey into super-resolution: A survey

    Saeed Anwar, Salman Khan, and Nick Barnes. A deep journey into super-resolution: A survey. ACM Computing Surveys, 2020. 1

    work page 2020

  3. [3]

    Super-resolution in medical imaging

    Greenspan, Hayit, 2009. Super-resolution in medical imaging. Comput. J. 52 (1), 43–63

    work page 2009

  4. [4]

    Huang, Yawen, Shao, Ling, Frangi, Alejandro F.,

    work page

  5. [5]

    Convolutional neu- ral network super resolution for face recognition in surveillance monitoring

    Rasti, Pejman, Uiboupin, Tonis, Escalera, Sergio, Anbarjafari, Gholamreza, 2016. Convolutional neu- ral network super resolution for face recognition in surveillance monitoring. In: Articulated Motion and Deformable Objects: 9th International Conference, AMDO 2016, Palma de Mallorca, Spain, July 13-15, 2016, Proceedings 9. Springer, pp. 175–184

    work page 2016

  6. [6]

    Remote Sensing Image Super-Resolution via Mixed High-Order Attention Network,

    D. Zhang, J. Shao, X. Li and H. T. Shen, "Remote Sensing Image Super-Resolution via Mixed High-Order Attention Network," in IEEE Transactions on Geoscience and Remote Sensing, vol. 59, no. 6, pp. 5183-5196, June 2021, doi: 10.1109/TGRS.2020.3009918

    arXiv 2021

  7. [7]

    Second-order attention network for single image super-resolution

    Tao Dai, Jianrui Cai, Yongbing Zhang, Shu-Tao Xia, and Lei Zhang. Second-order attention network for single image super-resolution. In 2019 IEEE/CVF Conference on Computer Vision and Pattern Recog- nition (CVPR), 2020. 1, 2, 7, 8

    work page 2019

  8. [8]

    Image super-resolution using deep con- volutional networks

    Chao Dong, Chen Change Loy, Kaiming He, and Xiaoou Tang. Image super-resolution using deep con- volutional networks. IEEE Transactions on Pattern Analysis and Machine Intelligence, page 295–307,

    work page

Show all 55 references
  1. [9]

    Deep learning for single image super-resolution: A brief review,

    Yang, W., Zhang, X., Tian, Y ., Wang, W., Xue, J.- H., and Liao, Q., “Deep learning for single image super-resolution: A brief review,” IEEE Transactions on Multimedia 21(12), 3106–3121 (2019)

    2019

  2. [11]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. Advances in neural information processing systems, 30, 2017. 1, 2, 4

    2017

  3. [12]

    Swin transformer: Hierarchical vision transformer us- ing shifted windows

    Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin transformer: Hierarchical vision transformer us- ing shifted windows. In International Conference on Computer Vision (ICCV), pages 10012–10022. IEEE,

  4. [13]

    arXiv preprint arXiv:2303.09735(2023)

    Zhou, Y ., Li, Z., Guo, C.L., Bai, S., Cheng, M.M., Hou, Q.: Srformer: Permuted self-attention for single image super-resolution. arXiv preprint arXiv:2303.09735(2023). 11, 12, 13

    2023 arXiv

  5. [14]

    Transformer for single image super-resolution

    Zhisheng Lu, Juncheng Li, Hong Liu, Chaoyan Huang, Lin lin Zhang, and Tieyong Zeng. Transformer for single image super-resolution. In Computer Vision and Pattern Recognition (CVPR), pages 457–466. IEEE/CVF, 2022. 1, 3

    2022

  6. [15]

    Activating more pixels in image super resolution transformer

    Xiangyu Chen, Xintao Wang, Jiantao Zhou, Yu Qiao, and Chao Dong. Activating more pixels in image super resolution transformer. In Computer Vision and Pattern Recognition (CVPR), pages 22367–22377. IEEE/CVF, 2023. 1, 3, 4, 5, 6, 7

    2023

  7. [16]

    Image super-resolution using deep convo- lutional networks

    Chao Dong, Chen Change Loy, Kaiming He, and Xi- aoou Tang. Image super-resolution using deep convo- lutional networks. IEEE transactions on pattern analy- sis and machine intelligence (PAMI), 38(2):295–307,

  8. [17]

    Accelerating the super-resolution convolutional neural network

    Chao Dong, Chen Change Loy, and Xiaoou Tang. Accelerating the super-resolution convolutional neural network. In Eur. Conf. Comput. Vis., 2016. 2

    2016

  9. [18]

    W. Shi, J. Caballero, F. Huszar, ´ J. Totz, A.P. Aitken, R. Bishop, D. Rueckert, Z. Wang, Real-time single image and video super-resolution using an efficient subpixel convolutional neural network, in: Proceed- ings of the IEEE Conference on Computer Vision and Pattern Recogni...

    2016

  10. [19]

    Enhanced deep residual networks for single image super-resolution

    Bee Lim, Sanghyun Son, Heewon Kim, Seungjun Nah, and Kyoung Mu Lee. Enhanced deep residual networks for single image super-resolution. In 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2017. 1, 2, 7

    2017

  11. [20]

    Residual dense network for image super- resolution, 2018

    Yulun Zhang, Yapeng Tian, Yu Kong, Bineng Zhong, and Yun Fu. Residual dense network for image super- resolution, 2018. 2

    2018

  12. [21]

    Photo-realistic single Jianfeng Wu et al.: Preprint submitted to Elsevier Page 8 of 10 Jianfeng Wu et al

    Christian Ledig, Lucas Theis, Ferenc Huszar, Jose Caballero, Andrew Cunningham, Alejandro Acosta, Andrew Aitken, Alykhan Tejani, Johannes Totz, Ze- han Wang, and Wenzhe Shi. Photo-realistic single Jianfeng Wu et al.: Preprint submitted to Elsevier Page 8 of 10 Jianfeng Wu et a...

  13. [22]

    Photo realistic single image super- resolution using a generative adversarial network

    Christian Ledig, Lucas Theis, Ferenc Huszar, Jose Caballero, ´ Andrew Cunningham, Alejandro Acosta, Andrew Aitken,Alykhan Tejani, Johannes Totz, Ze- han Wang, et al. Photo realistic single image super- resolution using a generative adversarial network. In IEEE Conf. Comput. Vi...

    2017

  14. [23]

    Real-esrgan: Training real-world blind super- resolution with pure synthetic data

    Xintao Wang, Liangbin Xie, Chao Dong, and Ying Shan. Real-esrgan: Training real-world blind super- resolution with pure synthetic data. In Int. Conf. Com- put. Vis., 2021. 2, 9

    2021

  15. [24]

    Generative adversarial networks

    Ian J Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial networks. arXiv:1406.2661, 2014. 2

    2014 arXiv

  16. [25]

    Non-local neural networks

    Xiaolong Wang, Ross Girshick, Abhinav Gupta, and Kaiming He. Non-local neural networks. In Proceed- ings of the IEEE conference on computer vision and pattern recognition, pages 7794–7803, 2018. 2

    2018

  17. [26]

    Second-order attention network for single image super-resolution

    Tao Dai, Jianrui Cai, Yongbing Zhang, Shu-Tao Xia, and Lei Zhang. Second-order attention network for single image super-resolution. In Computer Vision and Pattern Recognition (CVPR), pages 11065–11074. IEEE/CVF, 2019. 1, 2, 7, 8

    2019

  18. [28]

    Training ´ data-efficient image transformers & distillation through attention

    Hugo Touvron, Matthieu Cord, Matthijs Douze, Fran- cisco Massa, Alexandre Sablayrolles, and Herve J ´ egou. Training ´ data-efficient image transformers & distillation through attention. arXiv:2012.12877,

    2012 arXiv

  19. [29]

    Segmenter: Transformer for seman- tic segmentation

    Robin Strudel, Ricardo Garcia, Ivan Laptev, and Cordelia Schmid. Segmenter: Transformer for seman- tic segmentation. In Int. Conf. Comput. Vis., 2021. 2

    2021

  20. [30]

    Segformer: Simple and efficient design for semantic segmentation with transformers

    Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anand- kumar, Jose M Alvarez, and Ping Luo. Segformer: Simple and efficient design for semantic segmentation with transformers. In Adv. Neural Inform. Process. Syst., 2021. 2, 3, 4

    2021

  21. [31]

    Pre-trained image process- ing transformer

    Hanting Chen, Yunhe Wang, Tianyu Guo, Chang Xu, Yiping Deng, Zhenhua Liu, Siwei Ma, Chunjing Xu, Chao Xu, and Wen Gao. Pre-trained image process- ing transformer. In IEEE Conf. Comput. Vis. Pattern Recog., 2021. 1, 2, 6, 7

    2021

  22. [32]

    Vqfr: Blind face restoration with vector-quantized dictionary and parallel decoder

    Yuchao Gu, Xintao Wang, Liangbin Xie, Chao Dong, Gen Li, Ying Shan, and Ming-Ming Cheng. Vqfr: Blind face restoration with vector-quantized dictionary and parallel decoder. In Eur. Conf. Comput. Vis., 2022. 2

    2022

  23. [33]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al. An image is worth 16x16 words: Transformers for image recognition at scale. In Int. Conf. Learn....

  24. [34]

    Pre-trained image processing transformer

    Hanting Chen, Yunhe Wang, Tianyu Guo, Chang Xu, Yiping Deng, Zhenhua Liu, Siwei Ma, Chunjing Xu, Chao Xu, and Wen Gao. Pre-trained image processing transformer. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 12299–12310, 2021. 1, 2, 5, 6

    2021

  25. [35]

    Swin transformer: Hierarchical vision transformer using shifted windows

    Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin transformer: Hierarchical vision transformer using shifted windows. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 10012–10022, 2021. 1, 2, 3, 4

    2021

  26. [36]

    Vrt: A video restoration transformer, 2022

    Jingyun Liang, Jiezhang Cao, Yuchen Fan, Kai Zhang, Rakesh Ranjan, Yawei Li, Radu Timofte, and Luc Van Gool. Vrt: A video restoration transformer, 2022. 2

    2022

  27. [37]

    On efficient transformer and image pre- training for low level vision, 2021

    Wenbo Li, Xin Lu, Jiangbo Lu, Xiangyu Zhang, and Jiaya Jia. On efficient transformer and image pre- training for low level vision, 2021. 1, 2, 5, 6, 7

    2021

  28. [38]

    Swinir: Image restoration using swin transformer

    Jingyun Liang, Jiezhang Cao, Guolei Sun, Kai Zhang, LucVan Gool, and Radu Timofte. Swinir: Image restoration using swin transformer. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021. 1, 2, 3

    2021

  29. [39]

    Image super-resolution us- ing very deep residual channel attention networks

    Yulun Zhang, Kunpeng Li, Kai Li, Lichen Wang, Bineng Zhong, and Yun Fu. Image super-resolution us- ing very deep residual channel attention networks. In Proceedings of the European conference on computer vision (ECCV), 2018. 1, 2, 3, 4, 6

    2018

  30. [40]

    Si, Y ., Xu, H., Zhu, X., Zhang, W., Dong, Y ., Chen, Y ., & Li, H. (2024). SCSA: Exploring the Synergistic Effects Between Spatial and Channel Attention. arXiv e-prints. https://doi.org/10.48550/arXiv.2407.05128

    2024 doi arXiv

  31. [41]

    Image super- resolution using very deep residual channel attention networks

    Zhang, Yulun, Li, Kunpeng, Li, Kai, Wang, Lichen, Zhong, Bineng, Fu, Yun, 2018a. Image super- resolution using very deep residual channel attention networks. In: Proceedings of the European Conference on Computer Vision. ECCV , pp. 286–301. Jianfeng Wu et al.: Preprint submitt...

  32. [42]

    ECA-net: Ef- ficient channel attention for deep convolutional neural networks

    Wang, Qilong, Wu, Banggu, Zhu, Pengfei, Li, Peihua, Zuo, Wangmeng, Hu, Qinghua, 2020b. ECA-net: Ef- ficient channel attention for deep convolutional neural networks. In: Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition. pp. 11534–11542

  33. [43]

    chal- lenge on single image super-resolution: Methods and results

    R Timofte, S Gu, J Wu, and L NTIRE Van Gool. chal- lenge on single image super-resolution: Methods and results. In Computer Vision and Pattern Recognition (CVPR), pages 18–22. IEEE/CVF, 2018. 6

    2018

  34. [44]

    Enhanced deep residual net- works for single image super-resolution

    Bee Lim, Sanghyun Son, Heewon Kim, Seungjun Nah, and Kyoung Mu Lee. Enhanced deep residual net- works for single image super-resolution. In Computer Vision and Pattern Recognition Workshops (CVPR- W), pages 136–144. IEEE/CVF, 2017. 2, 3, 6, 7, 8

    2017

  35. [45]

    Low-complexity single- image super-resolution based on nonnegative neighbor embedding

    Marco Bevilacqua, Aline Roumy, Christine Guillemot, and Marie Line Alberi-Morel. Low-complexity single- image super-resolution based on nonnegative neighbor embedding. 2012. 6

    2012

  36. [46]

    On single image scale-up using sparse-representations

    Roman Zeyde, Michael Elad, and Matan Protter. On single image scale-up using sparse-representations. In International Conference Curves and Surfaces, pages 711–730. Springer, 2012. 6

    2012

  37. [47]

    A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics

    David Martin, Charless Fowlkes, Doron Tal, and Ji- tendra Malik. A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics. In In- ternational Conference on Computer Vision (ICCV), pages 416–423. ...

    2001

  38. [48]

    Single image super-resolution from transformed self- exemplars

    Jia-Bin Huang, Abhishek Singh, and Narendra Ahuja. Single image super-resolution from transformed self- exemplars. In Computer Vision and Pattern Recogni- tion (CVPR), pages 5197–5206. IEEE/CVF, 2015. 6

    2015

  39. [49]

    Sketch-based manga retrieval using manga109 dataset

    Yusuke Matsui, Kota Ito, Yuji Aramaki, Azuma Fu- jimoto, Toru Ogawa, Toshihiko Yamasaki, and Kiy- oharu Aizawa. Sketch-based manga retrieval using manga109 dataset. Multimedia Tools and Applica- tions, 76:21811–21838, 2017. 6

    2017

  40. [50]

    Adam: A method for stochastic optimization

    Kingma, Diederik P., Ba, Jimmy, 2014. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980

    2014 arXiv

  41. [51]

    Residual dense network for image super-resolution

    Zhang, Y ., Tian, Y ., Kong, Y ., Zhong, B., Fu, Y ., 2018b. Residual dense network for image super-resolution. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. pp. 2472–2481

  42. [52]

    Second-order attention network for single image super-resolution

    Tao Dai, Jianrui Cai, Yongbing Zhang, Shu-Tao Xia, and Lei Zhang. Second-order attention network for single image super-resolution. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition (CVPR), 2019. 6, 7

    2019

  43. [53]

    Cross-scale internal graph neural network for image super-resolution

    Zhou, S., Zhang, J., Zuo, W., Loy, C.C., 2020. Cross-scale internal graph neural network for image super-resolution. Adv. Neural Inf. Process. Syst. 33, 3499–3509

    2020

  44. [54]

    Single image super-resolution via a holistic attention network

    Niu, B., Wen, W., Ren, W., Zhang, X., Yang, L., Wang, S., Zhang, K., Cao, X., Shen, H., 2020. Single image super-resolution via a holistic attention network. In: Computer Vision–ECCV 2020: 16th European Con- ference, Glasgow, UK, August 23–28, 2020, Proceed- ings, Part XII 16....

    2020

  45. [55]

    Image super- resolution with non-local sparse attention

    Mei, Y ., Fan, Y ., Zhou, Y ., 2021. Image super- resolution with non-local sparse attention. In: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 3517–3526

    2021

  46. [56]

    On efficient transformer-based im- age pre-training for low-level vision

    Li, W., Lu, X., Qian, S., Lu, J., Zhang, X., Jia, J., 2021a. On efficient transformer-based im- age pre-training for low-level vision. arXiv preprint arXiv:2112.10175. Jianfeng Wu et al.: Preprint submitted to Elsevier Page 10 of 10

    arXiv

  47. [2017]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition

    Simultaneous super-resolution and cross- modality synthesis of 3D medical images using weakly-supervised joint convolutional sparse coding. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. pp. 6070–60444479

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.