Attention-Aware Linear Depthwise Convolution for Single Image Super-Resolution

Cheolkon Jung; Gwanghuyn Yu; Jinyoung Kim; Seongmin Hwang

REVIEW 4 major objections 5 minor 24 references

Attention-Aware Linear Depthwise Convolution for Single Image Super-Resolution

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 Removing the ReLU between depthwise and pointwise stages and scoring filters by determinant lets a lightweight network match heavier super-resolution models.

desk verdict Plausible lightweight SR block, but the state-of-the-art claim rests on an unfair comparison and the paper's own ablations don't single out the determinant descriptor. read the letter →

arxiv 1908.02648 v3 pith:55DZGR2U submitted 2019-08-07 eess.IV cs.CV

Seongmin Hwang , Gwanghuyn Yu , Cheolkon Jung , Jinyoung Kim This is my paper · ORCID

classification eess.IVcs.CV

keywords singleimagesuper-resolutiondepthwiseseparableconvolutionlinearchannelattentiondeterminantfilterinterdependencylightweightnetwork

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

Single-image super-resolution aims to restore a high-resolution image from a low-resolution input, and deep CNN models do this well but at growing computational cost; this paper attacks both problems at once. It proposes ALDNet, in which each depthwise convolution is made "linear" by removing the ReLU between the depthwise and pointwise stages, and an attention branch scores each depthwise filter by its determinant and rescales the channel features. The authors build this block into a lightweight baseline, ALDSR, and report that it matches or exceeds published results of VDSR, LapSRN, MemNet, and IDN on mainstream benchmarks at 4× upscaling with about 731k parameters. They also drop the block into RDN to form ALD-RDN and show it outperforms the corresponding depthwise-separable RDN on every benchmark. If the claims hold, the practical payoff is that a compact SR network can preserve fine details, such as repetitive lattice patterns, without paying the parameter cost of heavy models.

What carries the argument

The load-bearing unit is the attention-aware linear depthwise convolution (ALD convolution). It combines a depthwise $3\times3$ convolution directly followed by a $1\times1$ pointwise convolution with no activation in between, plus a branch that computes $\det W_3^c$ for each filter, passes the $C$-dimensional descriptor vector through FC-ReLU-FC-Sigmoid with reduction ratio $r=16$, and rescales the feature map as $f_c'=s_c f_c$, followed by residual scaling $D=(1+s_c)f_c$. The determinant is the paper's chosen descriptor of filter shape: Sarrus' rule gives the $3\times3$ determinant from products of diagonal entries, so the attention is driven by interdependency among a filter's weights rather than by spatial pooling of feature maps. A local residual connection every two ALD convolutions and a bottleneck $1\times1$ fusion realize the block's memory mechanism at low parameter count.

What would settle it

Retrain VDSR, LapSRN, MemNet, and IDN on the same DIV2K 800 training images with the same 4× bicubic protocol, loss, and training scheme used for ALDSR, and compare PSNR/SSIM on the four benchmark sets; if ALDSR no longer matches or exceeds them, the paper's headline superiority claim is not established. The same experiment could also test linearity by inserting a ReLU between the depthwise and pointwise stages of ALD-RDN and checking whether Urban100 performance drops.

Watch

Extended reading notes

Core claim

The paper's central claim is that depthwise separable convolution—originally designed for image classification—can be made effective for single-image super-resolution through two modifications. The first is removing the nonlinearity between the depthwise spatial filter and the 1×1 pointwise filter, because at that intermediate stage the feature map reflects only spatial information and a ReLU there discards reconstruction details. The second is an attention branch that exploits the fact that each depthwise filter is a small square matrix: the determinant of each $3\times3$ filter is computed by Sarrus' rule, reduced by a gating network, and used to recalibrate the channel features, with a residual connection so repeated attention does not degrade performance. The paper reports that the resulting ALDSR matches or beats the published numbers of several heavier SR networks, and that when inserted into RDN the proposed block (ALD-RDN) beats a depthwise-separable RDN on all tested datasets, with the largest gain on Urban100, which contains repetitive structures.

Load-bearing premise

The comparison in Table I assumes the published PSNR/SSIM values of VDSR, LapSRN, MemNet, and IDN—trained on different, generally smaller datasets—can be directly compared with ALDSR's scores, which were produced on the 800-image DIV2K training set with a different protocol.

Editorial extensions

If this is right

Linear depthwise convolution alone improves super-resolution: LDW-RDN beats DW-RDN on Urban100 by 0.08 dB with the same parameter count.
Adding the attention branch improves it further: ALD-RDN beats DW-RDN on all five test sets, with a 0.17 dB gain on Urban100.
The gains are cheap: one ALDB has 41,472 parameters versus 1,363,968 for an RDB, so the proposed block is suited to compact SR networks.
Determinant-based attention yields the best SSIM among the tested descriptors on every benchmark except Set5 for ALD-RDN, indicating better structural reconstruction.
The clearest effects appear on repeated patterns and textures, as shown on Urban100 images where ALDSR and ALD-RDN reconstruct lattices more accurately than the compared methods.

Reading between the lines

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

Because the block is modular, the same linear-depthwise-plus-determinant design could plausibly transfer to other image restoration tasks such as denoising, deblurring, or JPEG artifact removal, where preserving fine details is also critical; the paper only evaluates SR.
The determinant descriptor is one member of a family of matrix descriptors; a natural next experiment is comparing trace, spectral norm, or singular values in the attention branch, since the paper's ablation compares only determinant against average and max pooling.
A practical consequence the authors leave implicit is that replacing standard convolutions with ALD convolutions should cut FLOPs roughly by the depthwise factor, making real-time or on-device 4× SR feasible; the parameter counts in Table III support this direction.

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.

Desk Editor's Note

read the letter

Short version: the core idea here—dropping the ReLU between depthwise and pointwise convolutions and using a filter-interdependency attention branch—is sensible and not something I've seen before. But the state-of-the-art claim in the conclusion rests on Table I, which compares ALDSR trained on DIV2K (800 images) against published numbers for VDSR, LapSRN, MemNet, and IDN that were typically trained on 291 images. That comparison can't carry the weight. The differences are around 0.1–0.3 dB, exactly the size of typical training-data effects.

What the paper does well: the internal ablations are fairly designed. Replacing standard convs in RDN with depthwise separable (DW-RDN), then linear depthwise (LDW-RDN), then adding the attention branch (ALD-RDN) under the same protocol shows small but consistent gains, with the biggest on Urban100 (26.06 to 26.14 to 26.23). That suggests the mechanism is doing something real. The parameter counts are also reported clearly, and Table III shows ALDB is much lighter than RDB.

Soft spots: the determinant descriptor doesn't actually win on PSNR in the paper's own Table IV—average and max pooling are often better. The authors acknowledge this and fall back on SSIM, which is fine, but it undercuts the 'determinant captures filter shape' story. Also, no error bars, no multiple seeds, so even the internal gains could be noise; 0.08 dB on Urban100 is small. The paper doesn't release code, and the exact architecture of ALDSR (number of blocks, channels) is only partially specified.

Overall, this is an empirical architecture paper with a plausible idea and insufficient evidence for its headline. The stress-test concern lands: the SOTA claim is not established. The right fix is retraining baselines under the same protocol, reporting variance, and releasing code. The paper deserves a serious referee—not a desk reject—because the linear depthwise idea is simple and potentially useful for mobile SR. But I wouldn't cite the state-of-the-art claim as it stands. Readers working on efficient SR or lightweight networks should look at the internal ablations, not Table I.

Referee Report

4 major / 5 minor

Summary. The paper proposes ALDNet, a super-resolution network built on attention-aware linear depthwise convolution. The linear depthwise convolution removes the ReLU between the depthwise and pointwise convolutions, and an attention branch computes the determinant of each 3x3 depthwise filter and uses a gating mechanism to recalibrate feature maps. The authors construct a lightweight baseline, ALDSR, and also adapt the Residual Dense Network into ALD-RDN by replacing standard convolutions with the proposed ALD convolutions. Experiments are reported on Set5, Set14, BSD100, Urban100, and the DIV2K validation set, with internal ablations comparing depthwise, linear depthwise, and attention-aware variants under a common training protocol.

Significance. If the results hold, the work offers a lightweight SR architecture with a novel filter-descriptor-based attention mechanism. The internal ablation design is a strength: DW-RDN, LDW-RDN, and ALD-RDN are trained under the same protocol, and Table III provides useful parameter counts for the building blocks. The central claim, however, is not established because the headline comparison in Table I is not protocol-matched, and the determinant descriptor is not consistently the best on PSNR according to the paper's own Table IV. The paper would be significantly strengthened by retraining baselines under the same protocol and by reporting repeated-run statistics.

major comments (4)

[IV-B, Table I] The state-of-the-art comparison in Table I is not protocol-matched: ALDSR is trained on the 800-image DIV2K training set, while VDSR, LapSRN, and MemNet are commonly trained on the much smaller 291-image dataset with different augmentation and training schedules. Since SR performance is highly sensitive to training data quantity, the reported PSNR gaps do not demonstrate that ALDSR surpasses these methods. No baseline is retrained under the authors' protocol, and no error bars or multiple-seed results are provided. To support the conclusion that ALDNet achieves state-of-the-art performance, the authors must either retrain the baselines under the same protocol or visibly restrict the state-of-the-art claim to the internally re-implemented comparisons.
[V Conclusion and IV-B, Table IV] The conclusion states that the determinant descriptor shows superior performance to average and max descriptors, but this is contradicted by Table IV. On PSNR, average pooling or max pooling beats or ties the determinant on most configurations: for ALD-RDN, max pooling gives 32.23 dB on Set5 versus 32.18 dB for determinant, average pooling gives 28.66 dB on Set14 versus 28.62 dB, and the two descriptors tie on B100 and Urban100. For ALDSR, determinant is slightly worse on Urban100 and DIV2K validation than average pooling. The discussion in Section IV-B acknowledges this, so the concluding claim should be tempered and should distinguish PSNR from SSIM.
[IV-A, IV-B, Tables II and V] The evidence for the proposed mechanisms is based on small PSNR differences without statistical evaluation. For example, LDW-RDN outperforms DW-RDN by 0.08 dB on Urban100, and ALD-RDN outperforms DW-RDN by 0.17 dB on Urban100, but no repeated runs or variance estimates are reported. These differences may be within run-to-run variability, especially for networks trained with Adam and random patch sampling. Because these internal ablations are the main support for the claim that linearity and attention-aware depthwise convolution help SR, the authors should report the mean and standard deviation over at least two or three random seeds, or otherwise demonstrate that the improvements are reproducible.
[II-B, Eq. (3)] The reduction ratio r in the attention branch is set to 16 with no sensitivity analysis. Since the authors emphasize the low computational cost of the proposed block, the effect of varying r on both performance and parameter count should be reported, or at least a brief justification should be given for choosing the SENet-inspired value.

minor comments (5)

[I] There are several typos in the introduction and elsewhere, including 'depthwsie', 'mentined ealier', 'dose not', and 'convoltuion'. These should be corrected.
[III-A] The manuscript uses 'ALDSR' and 'ALDNet' interchangeably. The relationship between the general ALDNet and the specific ALDSR baseline should be stated explicitly once in Section III.
[Fig. 6] The text says the pointwise convolution layer is omitted from Figure 6 for simplicity, but this makes the block diagram misleading because the ALD convolution is defined as depthwise plus pointwise. The figure should include the pointwise layer or clearly annotate where it is inserted.
[IV-B, Discussion] The discussion of why max pooling helps PSNR and determinant helps SSIM is speculative; a short quantitative connection, such as correlation of descriptor values with reconstruction error, would make the interpretation more credible.
[References] Reference [4] (EDSR) is cited for the L1 loss and ESPCN but the discussion of batch normalization removal also relies on it; the citation is appropriate but the sentence structure could be clearer.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper's load-bearing claims are empirical architecture ablations, with no step reducing to its own inputs or to self-citation.

full rationale

This is an empirical architecture paper. The proposed mechanism, linear depthwise convolution, is defined independently as removal of ReLU between depthwise and pointwise convolutions, and its benefit is tested by direct ablation (Table V: DW-RDN vs. LDW-RDN). The attention mechanism uses a determinant descriptor of the depthwise filter weights and then recalibrates the feature map; this is an architectural ingredient evaluated by ablations in Tables II and IV, not a quantity fitted to the claimed result. No parameter is fitted to a subset of data and then renamed as a prediction, and no uniqueness theorem or load-bearing self-citation is invoked. The inherited reduction ratio r=16 from SENet is a hyperparameter choice, but the reported outputs are not defined in terms of it, so it is not circular. The main validity concern is that Table I compares ALDSR, trained on DIV2K, against published baseline numbers from methods trained on smaller datasets; that is an experimental fairness issue, not a circularity issue. Since no step can be exhibited as reducing to its own inputs, the circularity score is 0.

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

No new physical entities are introduced. The method rests on two domain heuristics: the determinant of a filter weight matrix is a useful descriptor for attention, and removing ReLU between depthwise and pointwise convolution preserves SR-relevant information. Both are empirically asserted rather than derived. The reduction ratio r=16 and the unstated network depth/width are hand-set choices.

free parameters (2)

reduction ratio r = 16
Set by hand following SENet; no sensitivity analysis is provided in the paper.
network depth/width (number of ALDBs and feature channels) = not reported
The architecture description omits the exact block count and channel numbers; parameters in Table III imply a configuration but it is not stated, making replication ambiguous.

assumptions (3)

domain assumption Determinant of a 3x3 depthwise filter is a meaningful descriptor of filter importance for channel attention
Section II-B asserts determinant captures shape and interdependencies; this is a heuristic, not derived, and Table IV shows it does not consistently beat average or max descriptors.
domain assumption Removing ReLU between depthwise and pointwise convolutions preserves information needed for SR without losing representational capacity
Central motivation in Section II-A; only supported by one ablation pair (Table V) on one training setup.
standard math Sarrus rule for 3x3 determinants is valid and appropriate here
Used to compute det(W^c_3) in Section II-B; standard linear algebra.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Attention-Aware Linear Depthwise Convolution for Single Image Super-Resolution." pith.science (2026). https://pith.science/paper/55DZGR2U

@misc{pith2026190802648,
  author       = {Pith},
  title        = {Pith review of: Attention-Aware Linear Depthwise Convolution for Single Image Super-Resolution},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/55DZGR2U}},
  note         = {Machine review of arXiv:1908.02648}
}

read the original abstract

Although deep convolutional neural networks (CNNs) have obtained outstanding performance in image superresolution (SR), their computational cost increases geometrically as CNN models get deeper and wider. Meanwhile, the features of intermediate layers are treated equally across the channel, thus hindering the representational capability of CNNs. In this paper, we propose an attention-aware linear depthwise network to address the problems for single image SR, named ALDNet. Specifically, linear depthwise convolution allows CNN-based SR models to preserve useful information for reconstructing a super-resolved image while reducing computational burden. Furthermore, we design an attention-aware branch that enhances the representation ability of depthwise convolution layers by making full use of depthwise filter interdependency. Experiments on publicly available benchmark datasets show that ALDNet achieves superior performance to traditional depthwise separable convolutions in terms of quantitative measurements and visual quality.

Figures

Figures reproduced from arXiv: 1908.02648 by the authors.

**Figure 3.** Sarrus’ rule to compute determinant. maps are incomplete. However, if the network learns the importance of each depthwise filter and recalibrates the feature values by correlation of depthwise filters, the quality of the feature maps are improved, thus leading to a remarkable improvement of the network performance. Driven by the observations, we propose attention-aware depthwise convolution which enables a network t… view at source ↗

**Figure 4.** Structure of attention branch. filter is represented by Wc 3 =   w c 1,1 w c 1,2 w c 1,3 w c 2,1 w c 2,2 w c 2,3 w c 3,1 w c 3,2 w c 3,3   (2) where c stands for channel. The rule of Sarrus is a well-known method to compute determinant of a 3×3 matrix. As illustrated in [PITH_FULL_IMAGE:figures/full_fig_p003_4.png] view at source ↗

Figures from the paper (4 more)

**Figure 5.** Figure 5: Structure of the proposed ALDSR. described in [8], an adequate gating function can be served by the sigmoid function. s = σ(WI δ(WDz + bD) + bI ), (3) where σ and δ are sigmoid function and rectified linear unit (ReLU), respectively; WI and WD are the weight set of con…

**Figure 6.** Figure 6: Structure of the attention-aware linear depthwise [PITH_FULL_IMAGE:figures/full_fig_p004_6.png]

**Figure 7.** Figure 7: Visual comparison for 4× SR on Urban100 dataset. [3] J. Kim, J. Kwon Lee, and K. Mu Lee, “Accurate image super-resolution using very deep convolutional networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 1646– 1654. [4…

**Figure 8.** Figure 8: Visual comparison for 4× SR on Urban100 dataset. D. Rueckert, and Z. Wang, “Real-time single image and video superresolution using an efficient sub-pixel convolutional neural network,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

24 extracted references · 11 canonical work pages

[1]

An edge-guided image interpolation algorithm via directional ﬁltering and data fusion,

L. Zhang and X. Wu, “An edge-guided image interpolation algorithm via directional ﬁltering and data fusion,” IEEE transactions on Image Processing, vol. 15, no. 8, pp. 2226–2238, 2006

work page 2006
[2]

Learning a deep convolu- tional network for image super-resolution,

C. Dong, C. C. Loy, K. He, and X. Tang, “Learning a deep convolu- tional network for image super-resolution,” in European conference on computer vision . Springer, 2014, pp. 184–199. 7 img004 from Urban100 HR (PSNR/SSIM) LapSRN (22.41dB/0.7989) Bicubic (21.09dB/0.6792) IDN (22.27dB/0.7972) VDSR (22.42dB/0.7955) ALDSR (22.95dB/0.8201) img085 from Urban100 ...

work page 2014
[3]

Accurate image super-resolution using very deep convolutional networks,

J. Kim, J. Kwon Lee, and K. Mu Lee, “Accurate image super-resolution using very deep convolutional networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2016, pp. 1646– 1654

work page 2016
[4]

Enhanced deep residual networks for single image super-resolution,

B. Lim, S. Son, H. Kim, S. Nah, and K. Mu Lee, “Enhanced deep residual networks for single image super-resolution,” in Proceedings of the IEEE conference on computer vision and pattern recognition workshops, 2017, pp. 136–144

2017
[5]

Residual dense network for image super-resolution,

Y . Zhang, Y . Tian, Y . Kong, B. Zhong, and Y . Fu, “Residual dense network for image super-resolution,” in Proceedings of the IEEE Con- ference on Computer Vision and Pattern Recognition , 2018, pp. 2472– 2481

2018
[6]

Mobilenets: Efﬁcient convo- lutional neural networks for mobile vision applications,

A. G. Howard, M. Zhu, B. Chen, D. Kalenichenko, W. Wang, T. Weyand, M. Andreetto, and H. Adam, “Mobilenets: Efﬁcient convo- lutional neural networks for mobile vision applications,” arXiv preprint arXiv:1704.04861, 2017

arXiv 2017
[7]

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,” arXiv preprint arXiv:1502.03167, 2015

arXiv 2015
[8]

Squeeze-and-excitation networks,

J. Hu, L. Shen, and G. Sun, “Squeeze-and-excitation networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 7132–7141

2018

Show all 24 references

[9]

Image super- resolution using very deep residual channel attention networks,

Y . Zhang, K. Li, K. Li, L. Wang, B. Zhong, and Y . Fu, “Image super- resolution using very deep residual channel attention networks,” in Proceedings of the European Conference on Computer Vision (ECCV) , 2018, pp. 286–301

2018
[10]

Second-order attention network for single image super-resolution,

T. Dai, J. Cai, Y . Zhang, S.-T. Xia, and L. Zhang, “Second-order attention network for single image super-resolution,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2019, pp. 11 065–11 074

2019
[11]

Very deep convolutional networks for large-scale image recognition,

K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” arXiv preprint arXiv:1409.1556 , 2014

2014 arXiv
[12]

Residual attention network for image classiﬁcation,

F. Wang, M. Jiang, C. Qian, S. Yang, C. Li, H. Zhang, X. Wang, and X. Tang, “Residual attention network for image classiﬁcation,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017, pp. 3156–3164

2017
[13]

Deep residual learning for image recognition,

K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2016, pp. 770–778

2016
[14]

Real-time single image and video super- resolution using an efﬁcient sub-pixel convolutional neural network,

W. Shi, J. Caballero, F. Husz ´ar, J. Totz, A. P. Aitken, R. Bishop, 8 img047 from Urban100 HR (PSNR/SSIM) LDW-RDN (21.94dB/0.7713) Bicubic (20.01dB/0.6282) ALD-RDN(avr) (21.96dB/0.7747) RDN(re-im) (22.16dB/0.7827) ALD-RDN(max) (21.94dB/0.7744) DW-RDN (21.90dB/0.7653) ALD-RDN(...

2016
[15]

Deep laplacian pyramid networks for fast and accurate super-resolution,

W.-S. Lai, J.-B. Huang, N. Ahuja, and M.-H. Yang, “Deep laplacian pyramid networks for fast and accurate super-resolution,” inProceedings of the IEEE conference on computer vision and pattern recognition , 2017, pp. 624–632

2017
[16]

Memnet: A persistent memory network for image restoration,

Y . Tai, J. Yang, X. Liu, and C. Xu, “Memnet: A persistent memory network for image restoration,” in Proceedings of the IEEE international conference on computer vision , 2017, pp. 4539–4547

2017
[17]

Fast and accurate single image super- resolution via information distillation network,

Z. Hui, X. Wang, and X. Gao, “Fast and accurate single image super- resolution via information distillation network,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2018, pp. 723–731

2018
[18]

Ntire 2017 challenge on single image super-resolution: Methods and results,

R. Timofte, E. Agustsson, L. Van Gool, M.-H. Yang, and L. Zhang, “Ntire 2017 challenge on single image super-resolution: Methods and results,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops , 2017, pp. 114–125

2017
[19]

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

M. Bevilacqua, A. Roumy, C. Guillemot, and M. L. Alberi-Morel, “Low- complexity single-image super-resolution based on nonnegative neighbor embedding,” 2012

2012
[20]

On single image scale-up using sparse-representations,

R. Zeyde, M. Elad, and M. Protter, “On single image scale-up using sparse-representations,” in International conference on curves and sur- faces. Springer, 2010, pp. 711–730

2010
[21]

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

D. Martin, C. Fowlkes, D. Tal, J. Malik et al. , “A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics.” Iccv Vancouver:, 2001

2001
[22]

Single image super-resolution from transformed self-exemplars,

J.-B. Huang, A. Singh, and N. Ahuja, “Single image super-resolution from transformed self-exemplars,” in Proceedings of the IEEE Confer- ence on Computer Vision and Pattern Recognition, 2015, pp. 5197–5206

2015
[23]

Adam: A method for stochastic optimization,

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

2014 arXiv
[24]

Automatic differentiation in pytorch,

A. Paszke, S. Gross, S. Chintala, G. Chanan, E. Yang, Z. DeVito, Z. Lin, A. Desmaison, L. Antiga, and A. Lerer, “Automatic differentiation in pytorch,” 2017. Seongmin Hwang received the B.S. and M.S. de- grees in electronic engineering from Chonnam Na- tional University, South...

2017

Pith tools

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

[1] [1]

An edge-guided image interpolation algorithm via directional ﬁltering and data fusion,

L. Zhang and X. Wu, “An edge-guided image interpolation algorithm via directional ﬁltering and data fusion,” IEEE transactions on Image Processing, vol. 15, no. 8, pp. 2226–2238, 2006

work page 2006

[2] [2]

Learning a deep convolu- tional network for image super-resolution,

C. Dong, C. C. Loy, K. He, and X. Tang, “Learning a deep convolu- tional network for image super-resolution,” in European conference on computer vision . Springer, 2014, pp. 184–199. 7 img004 from Urban100 HR (PSNR/SSIM) LapSRN (22.41dB/0.7989) Bicubic (21.09dB/0.6792) IDN (22.27dB/0.7972) VDSR (22.42dB/0.7955) ALDSR (22.95dB/0.8201) img085 from Urban100 ...

work page 2014

[3] [3]

Accurate image super-resolution using very deep convolutional networks,

J. Kim, J. Kwon Lee, and K. Mu Lee, “Accurate image super-resolution using very deep convolutional networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2016, pp. 1646– 1654

work page 2016

[4] [4]

Enhanced deep residual networks for single image super-resolution,

B. Lim, S. Son, H. Kim, S. Nah, and K. Mu Lee, “Enhanced deep residual networks for single image super-resolution,” in Proceedings of the IEEE conference on computer vision and pattern recognition workshops, 2017, pp. 136–144

2017

[5] [5]

Residual dense network for image super-resolution,

Y . Zhang, Y . Tian, Y . Kong, B. Zhong, and Y . Fu, “Residual dense network for image super-resolution,” in Proceedings of the IEEE Con- ference on Computer Vision and Pattern Recognition , 2018, pp. 2472– 2481

2018

[6] [6]

Mobilenets: Efﬁcient convo- lutional neural networks for mobile vision applications,

A. G. Howard, M. Zhu, B. Chen, D. Kalenichenko, W. Wang, T. Weyand, M. Andreetto, and H. Adam, “Mobilenets: Efﬁcient convo- lutional neural networks for mobile vision applications,” arXiv preprint arXiv:1704.04861, 2017

arXiv 2017

[7] [7]

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,” arXiv preprint arXiv:1502.03167, 2015

arXiv 2015

[8] [8]

Squeeze-and-excitation networks,

J. Hu, L. Shen, and G. Sun, “Squeeze-and-excitation networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 7132–7141

2018