Pith. sign in

REVIEW 4 major objections 6 minor 3 cited by

Towards Lightweight Hyperspectral Image Super-Resolution with Depthwise Separable Dilated Convolutional Network

T0 review · 4 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read The paper claims a 0.96M-parameter depthwise separable dilated convolutional network can match or beat much larger hyperspectral super-resolution models, topping PaviaC at 2x, 4x, and 8x.

desk verdict A reasonable lightweight HSI SR architecture, but the PaviaC advantage likely comes from test-set hyperparameter tuning, so the core empirical claim isn't yet supported. read the letter →

arxiv 2505.00374 v1 pith:DIVNPYRK submitted 2025-05-01 eess.IV cs.CV

Usman Muhammad , Jorma Laaksonen , Lyudmila Mihaylova This is my paper · ORCID
classification eess.IVcs.CV
keywords hyperspectralimagingsuper-resolutiondepthwiseseparableconvolutiondilatedfusionlightweightmodelspectralanglelossbandgroupingremotesensing
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

Hyperspectral super-resolution is typically tackled with large neural networks or by fusing extra panchromatic or RGB data, both impractical in many field settings. The paper proposes DSDCN, a lightweight network built from depthwise separable convolutions, a dilated fusion block, and a custom loss that combines pixel error, spectral angle, and an L2 constraint. The aim is to show that a 0.96-million-parameter model can reconstruct high-resolution hyperspectral images with spectral fidelity, outperforming all compared methods on PaviaC at every tested scale and staying competitive on PaviaU. If correct, this makes high-quality hyperspectral SR feasible on devices with limited memory and compute.

What carries the argument

The core components are (1) depthwise separable convolution blocks that decompose each convolution into per-band depthwise filtering followed by 1x1 pointwise mixing, with residual connections; (2) a dilated fusion block that runs three parallel 3x3 convolutions with dilation rates 1, 2, 3, concatenates their outputs, and fuses them with a 1x1 convolution to capture multi-scale context; (3) overlapping band grouping that splits the spectral bands into groups of size 32 with overlap; and (4) a custom loss $L_{total} = L_{MSE} + 0.5\, L_{SAM} + 0.03\, L_{\ell^2}$ combining mean squared error, spectral angle mapper loss, and an L2 constraint. The dilated fusion block is the mechanism that integrates spatial and spectral features, and the loss is what drives spectral integrity.

What would settle it

Fix the training protocol so that no hyperparameter is selected using the PaviaC or PaviaU test patches (for example, reserve a validation patch from the training region), then re-run the 2x, 4x, and 8x comparisons. If DSDCN's MPSNR no longer exceeds CSSFENet and PDENet on PaviaC, the claimed advantage is an artifact of test-set tuning.

Watch

Extended reading notes

Core claim

The central claim is that DSDCN achieves state-of-the-art or competitive hyperspectral super-resolution while using far fewer parameters than its rivals. On PaviaC it reports the best MPSNR, MSSIM, and SAM at 2x, 4x, and 8x, with 2x results of 36.434 dB and 0.9578 MSSIM; on PaviaU it is competitive but not consistently best (for example, 2x MPSNR of 35.941 dB versus 35.924 dB for CSSFENet). The authors attribute this to the combination of depthwise separable convolutions, a dilated fusion block with dilation rates 1, 2, and 3, and a loss function that explicitly penalizes spectral angle, all of which preserve spatial and spectral structure without enlarging the model.

Load-bearing premise

The load-bearing premise is that the band-group size, loss weights (0.5 and 0.03), and early stopping were chosen on a validation split separate from the test patches; if they were tuned on the PaviaC test patch, the reported gains over older methods would be optimistic.

Editorial extensions

If this is right

  • Hyperspectral SR could be deployed on drones, satellites, and embedded cameras where memory and power are tight, since the model fits in under 1 million parameters.
  • The method removes the need for panchromatic or RGB fusion, making single-image hyperspectral SR practical when auxiliary images are unavailable.
  • The spectral-angle loss term could be reused in other hyperspectral reconstruction tasks, such as denoising or pansharpening, without architectural changes.
  • If the results transfer, the same architecture can serve as a strong lightweight baseline for future hyperspectral SR research.

Reading between the lines

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

  • A natural next check is to evaluate DSDCN on a third hyperspectral dataset with no hyperparameter re-tuning; that would test whether the PaviaC advantage is general or dataset-specific.
  • The dilated fusion block with rates 1, 2, 3 could be viewed as a spectral-context analogue of multi-scale feature extraction, and its effectiveness here hints that similar fusion blocks may help in other spectral reconstruction problems.
  • The reported gap between PaviaC and PaviaU performance suggests the architecture may be sensitive to scene content or band configuration, so readers should expect the advantage to vary across datasets.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes DSDCN, a lightweight depthwise separable dilated convolutional network for hyperspectral image super-resolution. The architecture uses band grouping, depthwise separable convolution blocks with residual connections, a dilated convolution fusion block, and a custom loss combining MSE, spectral angle loss, and an L2 norm term. Experiments on the PaviaC and PaviaU datasets at 2x, 4x, and 8x downsampling factors report 0.96M parameters and claim competitive performance, with the best PaviaC numbers among the compared methods. The authors state that source code is publicly available.

Significance. The parameter efficiency of the proposed model (0.96M parameters) is a potentially useful contribution for practical hyperspectral super-resolution, and the combination of depthwise separable convolutions and dilated fusion is a reasonable design direction for lightweight SR. The paper explicitly ships source code and uses standard benchmark protocol inherited from [20], which aids reproducibility. However, the experimental validation as presented does not support the central claim of state-of-the-art performance, because hyperparameters are selected on the test patch and the reported margins are comparable to the tuning sensitivity. If the authors re-run the experiments with a proper validation split and the results still hold, the contribution would be solid; at present the significance claim is not trustworthy.

major comments (4)
  1. [III-A, Table I, Table II] The experimental protocol does not separate hyperparameter selection from test evaluation. Section III-A specifies that the test set is a single 144x144 patch (bottom center of PaviaC, top left of PaviaU) and that early stopping is used, but no validation split is described. Table I is an ablation study on PaviaC (2x) that selects the band group size (32) and the custom-loss configuration on what appears to be the same test patch that later produces the PaviaC results in Table II. This is test-set tuning. The quantitative impact is large: the Table I group-size sweep spans MPSNR from 35.017 to 36.434 dB, while the 2x PaviaC margin over CSSFENet in Table II is only 0.912 dB; with group size 16, DSDCN would be 35.447 dB and would fall below CSSFENet's 35.522 dB. The reported best results are therefore not independent, and the comparison to prior methods as reported is unfairly biased. The authors should introduce a separate validation split (or cross-validation), select all hyperparameters on the validation set only, report test metrics for the fixed configuration, and include results over multiple training runs.
  2. [I and II-C] The description of the dilated convolution fusion block in the Introduction disagrees with the equations in Section II-C. The Introduction states that the fusion block consists of a dilated convolution (spatial branch) and a pointwise convolution (spectral branch) whose outputs are concatenated. Equations (5)-(7), however, define three parallel 3x3 dilated convolutions with rates {1,2,3}, concatenation of their outputs, and a final 1x1 convolution; no separate pointwise spectral branch appears. The authors need to reconcile the text and the equations and specify which architecture is actually implemented in the results.
  3. [II-E, Eq. (12)] The third loss term duplicates the MSE term. The L2 loss in Eq. (12), L_ell2 = (1/N) sum_i ||y_true - y_pred||_2^2, is mathematically identical (up to a constant factor) to the MSE term, since both compute the mean per-pixel squared Euclidean difference across spectral bands. As a result, the proposed 'three-component' loss is effectively (1 + lambda2) times MSE plus lambda1 times SAM, and the claimed L2 norm constraint contributes no distinct objective. The authors should either remove the redundant term or replace it with a genuinely different regularizer, such as a spectral-gradient or band-wise consistency constraint, and update the ablation discussion accordingly.
  4. [Table II, PaviaU rows] The claim in the Abstract and Section III-C that the model achieves 'very competitive performance' on both datasets is not supported by the PaviaU results. On PaviaU at 2x, DSDCN has lower MSSIM (0.9442 vs 0.9625) and higher SAM (3.703 vs 3.038) than CSSFENet; at 4x and 8x, MPSNR is lower or equal, and MSSIM remains consistently lower. Only the MPSNR at 2x (35.941 vs 35.924) and SAM at 4x/8x are marginally favorable. The claims should be restricted to PaviaC or the method should be improved to actually compete on PaviaU.
minor comments (6)
  1. [Table I] The row labels 'DSDCN w/s 16' and 'DSDCN w/s 48' are unclear; 'w/s' should be spelled out as 'with group size' in the table caption or the label.
  2. [III-A] The overlap between band groups is described as 'one-fourth overlap', but it is not clear one-fourth of what (the group size, the total bands, or the stride). Please define the grouping and overlap precisely.
  3. [III-A] No error bars, standard deviations, or number of training seeds are reported for any of the metrics. Given the small test patch and the sensitivity shown in Table I, at least three seeds should be run and the mean plus standard deviation reported.
  4. [Table I] The parameter count comparison in Table I is reported only for scale 4; please state whether the parameter count is identical for 2x and 8x, or provide the counts per scale.
  5. [II-E] Referring to the L2 term as a 'constraint' is misleading because it is an unconstrained additive loss component; consider renaming it 'L2 regularization loss' or similar.
  6. [Section II] There is a typographical error in the contribution list: 'spatial resolution and preserve spectral integrity..' has a double period.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: the empirical architecture is evaluated against external baselines, with only a non-load-bearing self-citation for the SAM metric.

full rationale

The paper does not derive any claimed result from its own assumptions by definition, nor does it fit a parameter and then rename that fit a prediction. DSDCN is an empirical convolutional architecture whose reported MPSNR, MSSIM, and SAM values are compared with baseline numbers quoted from prior work [20], so the central evaluation is externally anchored rather than circular. The only self-citation among the authors is [36], used solely to name the Spectral Angle Mapper metric, and it is not load-bearing. The ablation in Table I selects band-group size and loss weights, and the experimental section states that the test patch is extracted from a fixed region with the remaining image regions used for training; if that ablation used the same test patch, this is a test-set hyperparameter selection concern that could bias the comparison, but it is a soundness issue rather than a circularity of the kind where a prediction reduces by construction to its input. The inclusion of an L2 term identical in form to the MSE term is redundant but does not constitute circular reasoning. Therefore no significant circularity is present; the slight score reflects only the minor self-citation, which has no load-bearing role.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim is an empirical performance and efficiency claim. It rests on a synthetic degradation model, an unstated band-group merge, hand-picked hyperparameters, and a small benchmark protocol drawn from [20].

free parameters (5)
  • SAM loss weight lambda1 = 0.5
    Chosen by hand in Eq. (12); no sensitivity analysis is reported, and results depend on it.
  • L2 loss weight lambda2 = 0.03
    Chosen by hand in Eq. (12); no sensitivity analysis is reported.
  • Band group size = 32
    Selected via ablation on PaviaC 2x in Table I; this is a data-dependent design choice.
  • Number of depthwise separable blocks = 3
    The paper says "empirically employ three such blocks" in Section II.B, with no ablation or justification.
  • Dilation rates = {1, 2, 3}
    Chosen for the three parallel branches in Section II.C; the paper provides no analysis of alternative rates.
assumptions (4)
  • domain assumption Area-based downsampling produces realistic LR-HR pairs for training and evaluation.
    Section III.A generates LR images by area-based sampling; the model is never tested on real sensor degradation, so this assumption is load-bearing for the practical claim.
  • domain assumption Band grouping into overlapping groups of 32 preserves all information needed for SR, and the unstated merge of group outputs is lossless.
    Section II.A introduces band grouping but does not describe how the reconstructed full image is assembled from overlapping subgroups, so correctness of the final output depends on this unstated step.
  • ad hoc to paper Three parallel dilated convolutions with rates {1, 2, 3} and concatenation suffice to capture the multi-scale spatial and spectral features needed.
    Section II.C defines this specific fusion block; no architectural search or ablation supports the choice.
  • ad hoc to paper The MSE plus SAM plus L2 loss with the chosen weights improves reconstruction without distorting optimization.
    Section II.E introduces the custom loss and lambda values; the only evidence is a single ablation row in Table I comparing with and without the custom loss.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards Lightweight Hyperspectral Image Super-Resolution with Depthwise Separable Dilated Convolutional Network." pith.science (2026). https://pith.science/paper/DIVNPYRK

@misc{pith2026250500374,
  author       = {Pith},
  title        = {Pith review of: Towards Lightweight Hyperspectral Image Super-Resolution with Depthwise Separable Dilated Convolutional Network},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DIVNPYRK}},
  note         = {Machine review of arXiv:2505.00374}
}
read the original abstract

Deep neural networks have demonstrated highly competitive performance in super-resolution (SR) for natural images by learning mappings from low-resolution (LR) to high-resolution (HR) images. However, hyperspectral super-resolution remains an ill-posed problem due to the high spectral dimensionality of the data and the scarcity of available training samples. Moreover, existing methods often rely on large models with a high number of parameters or require the fusion with panchromatic or RGB images, both of which are often impractical in real-world scenarios. Inspired by the MobileNet architecture, we introduce a lightweight depthwise separable dilated convolutional network (DSDCN) to address the aforementioned challenges. Specifically, our model leverages multiple depthwise separable convolutions, similar to the MobileNet architecture, and further incorporates a dilated convolution fusion block to make the model more flexible for the extraction of both spatial and spectral features. In addition, we propose a custom loss function that combines mean squared error (MSE), an L2 norm regularization-based constraint, and a spectral angle-based loss, ensuring the preservation of both spectral and spatial details. The proposed model achieves very competitive performance on two publicly available hyperspectral datasets, making it well-suited for hyperspectral image super-resolution tasks. The source codes are publicly available at: \href{https://github.com/Usman1021/lightweight}{https://github.com/Usman1021/lightweight}.

Figures

Figures reproduced from arXiv: 2505.00374 by the authors.

Figure 1
Figure 1. A detailed overview of the proposed model. Shallow features are initially extracted using a lightweight separable convolutional layer with ReLU [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

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

  1. Johnson-Lindenstrauss Lemma Guided Network for Efficient 3D Medical Segmentation

    cs.CV 2025-09 conditional novelty 5.0 of 10

    VeloxSeg pairs multi-scale window attention with JL-lemma-motivated grouped convolutions and Gram-matrix distillation to achieve efficient 3D medical segmentation at 1.66M parameters.

  2. A Fusion-Guided Inception Network for Hyperspectral Image Super-Resolution

    cs.CV 2025-05 conditional novelty 5.0 of 10

    FGIN is a 1.07M-parameter single-image hyperspectral super-resolution network that reports top MPSNR on PaviaC and competitive scores on PaviaU at 2x, 4x, and 8x scales.

  3. DACN: Dual-Attention Convolutional Network for Hyperspectral Image Super-Resolution

    eess.IV 2025-06 conditional novelty 4.0 of 10

    A dual-attention CNN with multi-head self-attention and channel attention improves hyperspectral super-resolution on PaviaC/PaviaU.

Reference graph

Works this paper leans on

36 extracted references · 18 canonical work pages · cited by 3 Pith papers

  1. [20]

    Hyperspectral image super-resolution based on feature diversity extraction,

    J. Zhang, R. Zheng, Z. Wan, R. Geng, Y . Wang, Y . Yang, X. Zhang, and Y . Li, “Hyperspectral image super-resolution based on feature diversity extraction,” Remote Sensing, vol. 16, no. 3, p. 436, 2024

    2024

  2. [1]

    Hyperspectral remote sensing in precision agriculture: Present status, challenges, and future trends,

    P. Singh, P. C. Pandey, G. P. Petropoulos, A. Pavlides, P. K. Srivastava, N. Koutsias, K. A. K. Deng, and Y . Bao, “Hyperspectral remote sensing in precision agriculture: Present status, challenges, and future trends,” in Hyperspectral remote sensing . Elsevier, 2020, pp. 121–146

    work page 2020

  3. [2]

    Hyperspectral pansharpening: A review,

    L. Loncan, L. B. de Almeida, J. M. Bioucas-Dias, X. Briottet, J. Chanus- sot, N. Dobigeon, S. Fabre, W. Liao, G. A. Licciardi, M. Sim ˜oes, J.- Y . Tourneret, M. A. Veganzones, G. Vivone, Q. Wei, and N. Yokoya, “Hyperspectral pansharpening: A review,” IEEE Geoscience and Remote Sensing Magazine, vol. 3, no. 3, pp. 27–46, 2015

    work page 2015

  4. [3]

    Supervised nonlinear spectral unmixing using a postnonlinear mixing model for hyperspectral imagery,

    Y . Altmann, A. Halimi, N. Dobigeon, and J.-Y . Tourneret, “Supervised nonlinear spectral unmixing using a postnonlinear mixing model for hyperspectral imagery,” IEEE Transactions on Image Processing, vol. 21, no. 6, pp. 3017–3025, 2012

    work page 2012

  5. [4]

    Spectral variability in hyperspectral data unmixing: A comprehensive review,

    R. A. Borsoi, T. Imbiriba, J. C. M. Bermudez, C. Richard, J. Chanussot, L. Drumetz, J.-Y . Tourneret, A. Zare, and C. Jutten, “Spectral variability in hyperspectral data unmixing: A comprehensive review,” IEEE Geo- science and Remote Sensing Magazine , vol. 9, no. 4, pp. 223–270, 2021

    work page 2021

  6. [5]

    Pre-trained vggnet architecture for remote-sensing image scene classification,

    U. Muhammad, W. Wang, S. P. Chattha, and S. Ali, “Pre-trained vggnet architecture for remote-sensing image scene classification,” in 2018 24th International Conference on Pattern Recognition (ICPR) . IEEE, 2018, pp. 1622–1627

    2018

  7. [6]

    Feature fusion with deep supervision for remote-sensing image scene classification,

    U. Muhammad, W. Wang, and A. Hadid, “Feature fusion with deep supervision for remote-sensing image scene classification,” in 2018 IEEE 30th international conference on tools with artificial intelligence (ICTAI). IEEE, 2018, pp. 249–253

    2018

  8. [7]

    Bag of words kaze (bowk) with two-step classification for high-resolution remote sensing images,

    U. Muhammad, W. Wang, A. Hadid, and S. Pervez, “Bag of words kaze (bowk) with two-step classification for high-resolution remote sensing images,” IET Computer Vision , vol. 13, no. 4, pp. 395–403, 2019

    2019

Show all 36 references
  1. [8]

    Patch-based discriminative learning for remote sensing scene classification,

    U. Muhammad, M. Z. Hoque, W. Wang, and M. Oussalah, “Patch-based discriminative learning for remote sensing scene classification,” Remote Sensing, vol. 14, no. 23, p. 5913, 2022

    2022

  2. [9]

    Lcrca: image super- resolution using lightweight concatenated residual channel attention net- works,

    C. Peng, P. Shu, X. Huang, Z. Fu, and X. Li, “Lcrca: image super- resolution using lightweight concatenated residual channel attention net- works,” Applied Intelligence, pp. 1–15, 2022

    2022

  3. [10]

    Sbhsr: single-band super- resolution method for hyperspectral images based on blind degradation and fusion of auxiliary band,

    L. Bu, X. Xiao, Z. Zhang, and D. Mingjun, “Sbhsr: single-band super- resolution method for hyperspectral images based on blind degradation and fusion of auxiliary band,” Geocarto International, vol. 39, no. 1, p. 2294901, 2024

    2024

  4. [11]

    Deep posterior distribution-based embedding for hyperspectral image super-resolution,

    J. Hou, Z. Zhu, J. Hou, H. Zeng, J. Wu, and J. Zhou, “Deep posterior distribution-based embedding for hyperspectral image super-resolution,” IEEE Transactions on Image Processing , vol. 31, pp. 5720–5732, 2022

    2022

  5. [12]

    Msdformer: Multi-scale deformable transformer for hyperspectral image super-resolution,

    S. Chen, L. Zhang, and L. Zhang, “Msdformer: Multi-scale deformable transformer for hyperspectral image super-resolution,” IEEE Transactions on Geoscience and Remote Sensing , 2023

    2023

  6. [13]

    Towards lightweight super-resolution with dual regression learning,

    Y . Guo, J. Wang, Q. Chen, J. Cao, Z. Deng, Y . Xu, J. Chen, and M. Tan, “Towards lightweight super-resolution with dual regression learning,” IEEE Transactions on Pattern Analysis and Machine Intelligence , 2024

    2024

  7. [14]

    Learning with privileged infor- mation for efficient image super-resolution,

    W. Lee, J. Lee, D. Kim, and B. Ham, “Learning with privileged infor- mation for efficient image super-resolution,” in Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XXIV 16 . Springer, 2020, pp. 465–482

    2020

  8. [15]

    Image super-resolution us- ing knowledge distillation,

    Q. Gao, Y . Zhao, G. Li, and T. Tong, “Image super-resolution us- ing knowledge distillation,” in Asian Conference on Computer Vision . Springer, 2018, pp. 527–541

    2018

  9. [16]

    Achieving on-mobile real- time super-resolution with neural architecture and pruning search,

    Z. Zhan, Y . Gong, P. Zhao, G. Yuan, W. Niu, Y . Wu, T. Zhang, M. Jayaweera, D. Kaeli, B. Ren et al. , “Achieving on-mobile real- time super-resolution with neural architecture and pruning search,” in Proceedings of the IEEE/CVF international conference on computer vision, 202...

    2021

  10. [17]

    A convex low-rank regularization method for hyperspectral super-resolution,

    R. Wu, Q. Li, X. Fu, and W.-K. Ma, “A convex low-rank regularization method for hyperspectral super-resolution,” in 2018 IEEE Statistical Signal Processing Workshop (SSP) . IEEE, 2018, pp. 383–387

    2018

  11. [18]

    Binarized neural network for single image super resolution,

    J. Xin, N. Wang, X. Jiang, J. Li, H. Huang, and X. Gao, “Binarized neural network for single image super resolution,” in Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part IV 16 . Springer, 2020, pp. 91–107

    2020

  12. [19]

    Efficient super resolution using binarized neural network,

    Y . Ma, H. Xiong, Z. Hu, and L. Ma, “Efficient super resolution using binarized neural network,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops , 2019, pp. 0–0

    2019

  13. [21]

    Mobilenets: Efficient convolutional neural networks for mobile vision applications,

    A. G. Howard, “Mobilenets: Efficient convolutional neural networks for mobile vision applications,” arXiv preprint arXiv:1704.04861 , 2017

    2017 arXiv

  14. [22]

    Fast, accurate, and lightweight super-resolution with cascading residual network,

    N. Ahn, B. Kang, and K.-A. Sohn, “Fast, accurate, and lightweight super-resolution with cascading residual network,” in Proceedings of the European conference on computer vision (ECCV) , 2018, pp. 252–268

    2018

  15. [23]

    Deeply-recursive convolutional net- work for image super-resolution,

    J. Kim, J. K. Lee, and K. M. Lee, “Deeply-recursive convolutional net- work for image super-resolution,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2016, pp. 1637–1645

    2016

  16. [24]

    Hdconv: Heterogeneous kernel-based dilated convolutions,

    H. Hu, C. Yu, Q. Zhou, Q. Guan, and H. Feng, “Hdconv: Heterogeneous kernel-based dilated convolutions,” Neural Networks, vol. 179, p. 106568, 2024

    2024

  17. [25]

    A lightweight neural network combining dilated convolution and depthwise separable convolution,

    W. Sun, X. Zhou, X. Zhang, and X. He, “A lightweight neural network combining dilated convolution and depthwise separable convolution,” in Cloud Computing, Smart Grid and Innovative Frontiers in Telecommuni- cations: 9th EAI International Conference, CloudComp 2019, and 4th EA...

    2019

  18. [26]

    Multi-scale context aggregation by dilated convo- lutions,

    F. Yu and V . Koltun, “Multi-scale context aggregation by dilated convo- lutions,” arXiv preprint arXiv:1511.07122 , 2015

    2015 arXiv

  19. [27]

    Enhancing hy- perspectral images via diffusion model and group-autoencoder super- resolution network,

    Z. Wang, D. Li, M. Zhang, H. Luo, and M. Gong, “Enhancing hy- perspectral images via diffusion model and group-autoencoder super- resolution network,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 38, no. 6, 2024, pp. 5794–5804

    2024

  20. [28]

    Mobilenetv2: Inverted residuals and linear bottlenecks,

    M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L.-C. Chen, “Mobilenetv2: Inverted residuals and linear bottlenecks,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 4510–4520

    2018

  21. [29]

    Accelerating the super-resolution convolutional neural network,

    C. Dong, C. C. Loy, and X. Tang, “Accelerating the super-resolution convolutional neural network,” in Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part II 14 . Springer, 2016, pp. 391–407

    2016

  22. [30]

    Exploring the relationship between 2d/3d convolution for hyperspectral image super-resolution,

    Q. Li, Q. Wang, and X. Li, “Exploring the relationship between 2d/3d convolution for hyperspectral image super-resolution,” IEEE Transactions on Geoscience and Remote Sensing, vol. 59, no. 10, pp. 8693–8703, 2021

    2021

  23. [31]

    Mixed 2d/3d convolutional network for hyperspectral image super-resolution,

    Q. W. Qiang Li and X. Li, “Mixed 2d/3d convolutional network for hyperspectral image super-resolution,” Remote sensing , vol. 12, no. 10, p. 1660, 2020

    2020

  24. [32]

    Accurate image super-resolution using very deep convolutional networks,

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

    2016

  25. [33]

    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

  26. [34]

    Multi-scale feature mapping net- work for hyperspectral image super-resolution,

    J. Zhang, M. Shao, Z. Wan, and Y . Li, “Multi-scale feature mapping net- work for hyperspectral image super-resolution,” Remote Sensing, vol. 13, no. 20, p. 4180, 2021

    2021

  27. [35]

    As 3 itransunet: Spatial-spectral interactive transformer u-net with alternating sampling for hyperspectral image super-resolution,

    Q. Xu, S. Liu, J. Wang, B. Jiang, and J. Tang, “As 3 itransunet: Spatial-spectral interactive transformer u-net with alternating sampling for hyperspectral image super-resolution,” IEEE Transactions on Geoscience and Remote Sensing , 2023

    2023

  28. [36]

    A comparison of hyperspectral super-resolution techniques for boreal forest imagery,

    Y . Chudasama, U. Muhammad, V . Mayra, F. Guiotte, and J. Laaksonen, “A comparison of hyperspectral super-resolution techniques for boreal forest imagery,” in IGARSS 2024-2024 IEEE International Geoscience and Remote Sensing Symposium . IEEE, 2024, pp. 1226–1230

    2024

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.