Pith. sign in

REVIEW 3 major objections 7 minor 43 references

Beyond Pretty Pictures: Combined Single- and Multi-Image Super-resolution for Sentinel-2 Images

T0 review · 3 major / 7 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 A hybrid network that combines repeated Sentinel-2 passes with a learned high-resolution prior produces super-resolved images that beat both pure single-image and pure multi-image methods on land-cover classification.

desk verdict Careful SR evaluation with a real capacity confound; worth refereeing, but the hybrid claim needs a same-size SISR control. read the letter →

arxiv 2505.24799 v2 pith:NTU75VU4 submitted 2025-05-30 eess.IV cs.CV

Aditya Retnanto (1) , Son Le (1) , Sebastian Mueller (1) , Armin Leitner (2) , Michael Riffler (2) , Konrad Schindler (3) , Yohan Iddawela (1) ((1) Asian Development Bank , Philippines
show 5 more authors
(2) GeoVille Information Systems Data Processing GmbH Austria (3) ETH Zürich Switzerland)
This is my paper
classification eess.IVcs.CV
keywords super-resolutionSentinel-2multi-imagesingle-imageland-coverclassificationremotesensinghybridfusionPléiadesNeo
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 argues that super-resolution of freely available Sentinel-2 imagery can be made genuinely useful for mapping if one combines two complementary mechanisms in a single network: multi-image fusion that exploits subtle shifts across repeated acquisitions, and a single-image learned prior that supplies fine detail the fusion misses. The authors build SEN4X, which turns 10 m Sentinel-2 images into 2.5 m images, and evaluate the result by feeding it to a land-cover classifier rather than by measuring pixel similarity. On a Hanoi test set, SEN4X reaches 74.6% overall accuracy and 51.6% mean IoU, ahead of the pure SISR baseline Swin2SR (71.4%, 48.9%) and the pure MISR baseline HighResNet (58.3%, 38.7%). The same evaluation shows PSNR and SSIM rank methods differently than the downstream task, so tuning SR on those metrics can select models that do not actually help mapping.

What carries the argument

The load-bearing mechanism is early fusion with a deferred strong prior: eight low-resolution views are each passed through a 3x3 shallow feature extractor, recursively merged in pairs by HighResNet-style residual fusion blocks, and only then processed by six residual Swin transformer blocks and a 4x pixel-shuffle upsampler. This ordering matters: the late-fusion variant SEN4X[late] performs worse and is roughly eight times slower because the Swin backbone runs per view. The learned prior comes from training on real Sentinel-2/Pléiades Neo pairs after per-tile histogram matching, so the network learns a mapping from actual 10 m reflectance to actual 2.5 m detail instead of from synthetic downsampling.

What would settle it

Run the identical SEN4X pipeline on a geographically separate city with its own Pléiades Neo targets; if the mIoU advantage over Swin2SR falls to near zero or per-class gains invert, then the Hanoi result is explained by dataset-specific texture rather than by the fusion mechanism.

Watch

Extended reading notes

Core claim

On its own terms, the discovery is that hybrid SR is better than either branch alone for downstream land-cover segmentation. SEN4X first runs eight co-registered Sentinel-2 revisits through a shallow feature extractor, recursively fuses the resulting feature maps with the pairwise residual fusion blocks of HighResNet, then refines the single fused representation with six residual Swin transformer blocks and upsamples 4x with pixel shuffle. Trained end-to-end on real cross-sensor pairs (Sentinel-2 low-resolution, Pléiades Neo high-resolution) and tested by classifying urban land cover, it improves mIoU by 2.7 points over Swin2SR and 12.9 points over HighResNet. The paper further claims that conventional image-quality metrics, especially PSNR and hallucination scores, are poor predictors of this downstream utility, while LPIPS agrees with segmentation rankings.

Load-bearing premise

The claim depends on the assumption that the high-resolution Pléiades Neo images, after simple per-tile brightness adjustment, are trustworthy stand-ins for what Sentinel-2 would see at 2.5 m; if the two sensors' color and illumination differences remain, the network may be learning to paint in details that are not real.

Editorial extensions

If this is right

  • If correct, free Sentinel-2 time series can partially substitute expensive high-resolution imagery for urban land-cover mapping, narrowing the accuracy gap from 41.6 points for bicubic upsampling to 11.0 points relative to true Pléiades Neo data.
  • Super-resolution is more useful as a separate preprocessing step than as an end in itself, because one enhanced image can serve multiple downstream tasks without per-task high-resolution guidance.
  • When repeated acquisitions exist, hybrid SR should be preferred over pure SISR or pure MISR: the fusion adds detail while the learned prior adds sharpness, and the mIoU gains are 2.7 and 12.9 points respectively.
  • Evaluation of SR should include a downstream task, because PSNR and SSIM rank HighResNet favorably despite its poor segmentation, while LPIPS closely tracks land-cover classification utility.
  • The fusion order matters more than model capacity alone, since SEN4X[late] with the same parameter count performs worse than the default early-fusion design and costs about eight times more compute.

Reading between the lines

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

  • The 2.3-point mIoU advantage over ESRGAN suggests that a dedicated recursive fusion front end extracts more from the same eight views than simple input stacking; an ablation that swaps SEN4X's fusion module for stacked inputs would test this directly.
  • If the approach transfers geographically, the same hybrid design could be applied to Sentinel-2's red-edge and shortwave-infrared bands once adequate high-resolution references exist, or to other free moderate-resolution sensors with dense time series.
  • Because the evaluation is confined to Hanoi, part of the gain may come from memorized urban texture rather than the fusion mechanism; a second city with independent high-resolution targets would clarify the source of the improvement.
  • The paper's metric finding implies that published SR benchmarks reporting only PSNR and SSIM may not reflect practical value, so future comparison studies should re-rank those models on segmentation or other downstream tasks.
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

3 major / 7 minor

Summary. This paper proposes SEN4X, a hybrid single-image and multi-image super-resolution network for Sentinel-2 imagery. The network fuses eight temporally selected Sentinel-2 revisits with a Swin2SR-style learned prior and a HighResNet-style recursive fusion module, producing 2.5 m RGB+NIR images from 10 m input. Training targets are Pléiades Neo images after per-tile histogram matching. The authors evaluate the SR outputs by feeding them into a Satlas-based land-cover segmentation model trained on the same PNEO images, over a held-out geographic test set in Hanoi. They report that SEN4X reaches 74.6% accuracy and 51.6% mIoU, surpassing Swin2SR (71.4%, 48.9%), ESRGAN (72.4%, 49.3%), and HighResNet (58.3%, 38.7%), and that PSNR/SSIM do not track segmentation performance. All methods are trained with five random seeds under a consistent protocol.

Significance. The paper's evaluation design is a methodological strength: all methods are trained on the same data with consistent hyperparameters, evaluated on geographically held-out regions, repeated over five random seeds, and judged by a downstream land-cover classification task rather than by pixel metrics alone. The finding that conventional image-quality metrics misrank SR methods for a concrete Earth-observation task is a valuable message for the community. The promised public release of code and trained models supports reproducibility. However, the central scientific claim that the hybrid MISR+SISR combination is responsible for the improvement is not yet supported, because the comparison baselines are smaller than SEN4X and the provided ablation varies only the fusion order.

major comments (3)
  1. [§5.2, Table 4] The comparison that supports the central hybrid-fusion claim confounds architecture with model capacity. SEN4X has 30,517,135 parameters, whereas the SISR-only Swin2SR baseline has 24,525,082 and the hybrid ESRGAN baseline has 16,715,268. The text in §5.2 states that the SEN4X–Swin2SR gap 'is likely due to the inclusion of the MISR component,' but a conventional SISR model with the same capacity as SEN4X is not evaluated. The observed 2.7 pp mIoU advantage could therefore be a capacity effect rather than a fusion effect. Please add a same-capacity SISR-only control (e.g., the SEN4X backbone with the fusion branch removed, or a Swin2SR scaled to about 30M parameters) and, ideally, a same-capacity MISR-only control, and re-evaluate the attribution.
  2. [§4.1, Table 1] The ablation SEN4X[late] keeps both the SISR and MISR components and varies only the order of fusion; it does not test whether the multi-image branch contributes at all. Because the central claim is that combining the two components is beneficial, the paper needs an ablation that removes the MISR branch entirely (SISR only, same capacity) and one that removes the SISR branch (MISR only, same capacity). Without these controls, the hybrid advantage is not established beyond the existing baselines.
  3. [§3.2, Figure 1] The cross-sensor calibration step is central to the claim that SR improves physical realism, but the paper reports only a histogram of L1 differences before and after histogram matching and no quantitative residual misalignment per spectral band or per tile. Since the SR models and the land-cover classifier are both trained on the same PNEO images, the downstream evaluation could reward PNEO-specific artifacts rather than true surface-reflectance detail. Please report per-band residual statistics on the held-out test tiles (e.g., mean absolute error and bias after calibration) and, if possible, evaluate with an independent high-resolution label source to strengthen the physical-consistency claim.
minor comments (7)
  1. [Throughout] The model name is inconsistently capitalized as 'SEN4X' and 'Sen4x' across the text, tables, and figures; please unify the notation.
  2. [§4.1] The embedding dimension is stated as '258'; please verify whether this is intended or a typo for 256.
  3. [Table 4] The inference-time entries are formatted irregularly (e.g., '0133.6±.07.9'); please reformat the table and state the units consistently.
  4. [§5.2] The paper reports mean and standard deviation over five seeds but does not provide a significance test for the key mIoU differences; a paired test or confidence intervals would strengthen the claims.
  5. [§3.2] The per-tile histogram matching is described, but no quantitative residual error is reported beyond Figure 1; please add per-band mean absolute error and bias statistics on the held-out tiles to support the radiometric alignment claim.
  6. [Figure 1] The caption uses 'L1-differences'; please use 'L1 differences' and define the metric in the text.
  7. [§3.1] The notation '2.5km2' for tile area is ambiguous; please use '2.5 km²' and clarify the tile dimensions in meters (158 × 158 pixels at 10 m implies about 1.58 km × 1.58 km).

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central result is an empirical, held-out land-cover evaluation with a frozen downstream classifier, not a derivation that reduces to its inputs.

full rationale

The paper's central claim is an empirical ranking of super-resolution methods by a downstream land-cover segmentation task on a geographically held-out test set (Section 5.2, Table 1). The SR models are trained with a reconstruction objective against harmonized Pléiades Neo imagery, not against the reported mIoU or accuracy, and the segmentation model is trained once on the PNEO targets and then frozen; therefore the reported SEN4X ranking is not forced by the training losses. The shared use of PNEO imagery as both SR target and segmentation training data (Sections 3.2 and 4.3) is an upper-bound design choice, not a fitted-input prediction, since all SR baselines are trained against the same PNEO target yet produce significantly different classification scores. The paper's self-citations ([7], [22]) occur only in related-work context and are not load-bearing for SEN4X's performance. The parameter-count confound between SEN4X and the baseline architectures is a real experimental-design concern but is not circularity: it does not make the comparison true by construction. The paper also explicitly acknowledges its geographic and single-task limitations in Section 6, which further supports treating the findings as empirical rather than definitional. No circular step was found.

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

This is an empirical deep learning paper, so the ledger captures dataset construction assumptions and architecture hyperparameters rather than mathematical postulates. The central claim (SEN4X improves downstream classification) depends on the validity of the cross-sensor alignment, the quality of the manual labels, and the transferability of the segmentation model.

free parameters (4)
  • Number of input views = 8
    The MISR component uses eight Sentinel-2 revisits per tile, following the SATLAS finding. This is a hand-chosen hyperparameter that the performance depends on; using fewer or more views changes the oversampling and results.
  • Swin transformer embedding dimension = 258
    The SISR backbone uses a non-standard embedding dimension of 258 (modified from Swin2SR), chosen by the authors without a reported ablation. The capacity of the model depends on this.
  • Swin window size = 8
    Window size set to 8 for the multi-head self-attention in the RSTBs. This is an architecture hyperparameter affecting the receptive field.
  • Number of RSTB blocks = 6
    The deep feature extractor uses six residual Swin transformer blocks. The depth is a design choice that influences performance and compute.
assumptions (5)
  • domain assumption Sentinel-2 Level-2A surface reflectance and Pléiades Neo top-of-atmosphere reflectance can be radiometrically aligned via histogram matching such that SR training pairs are valid.
    Section 3.2: PNEO values are normalized and histogram-matched to the best LR image per tile. If this alignment is imperfect, the learned SR mapping may be biased.
  • domain assumption The eight selected LR revisits provide sufficient sub-pixel shifts and temporal consistency to act as oversampling for MISR.
    Section 3.1: The eight most suitable revisits are selected using temporal proximity, completeness, and spectral quality. The assumption is that these images contain complementary information and minimal land-cover change.
  • domain assumption The land-cover labels, manually annotated from PNEO imagery and augmented with Google Open Buildings, are accurate ground truth.
    Section 3.3: Labels are manually drawn and rasterized, retaining only homogeneous pixels. Label errors would directly propagate into the mIoU scores used to rank SR methods.
  • domain assumption The segmentation model trained on PNEO images at 2.5m is applicable to SR images from other sensors after harmonization.
    Section 4.3: The SATLAS foundation model is pretrained on 0.5-2.5m RGB data and fine-tuned on PNEO. The evaluation assumes that domain shift between PNEO and SR images does not invalidate the comparison, though the shift is likely similar across SR methods.
  • domain assumption The test regions in northern and eastern Hanoi are geographically distinct from training regions, so the reported performance is not due to spatial autocorrelation.
    Section 3.4: Two contiguous regions are set aside for testing. This assumes the stratification prevents leakage, which is plausible but not quantitatively verified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Beyond Pretty Pictures: Combined Single- and Multi-Image Super-resolution for Sentinel-2 Images." pith.science (2026). https://pith.science/paper/NTU75VU4

@misc{pith2026250524799,
  author       = {Pith},
  title        = {Pith review of: Beyond Pretty Pictures: Combined Single- and Multi-Image Super-resolution for Sentinel-2 Images},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NTU75VU4}},
  note         = {Machine review of arXiv:2505.24799}
}
read the original abstract

Super-resolution aims to increase the resolution of satellite images by reconstructing high-frequency details, which go beyond na\"ive upsampling. This has particular relevance for Earth observation missions like Sentinel-2, which offer frequent, regular coverage at no cost; but at coarse resolution. Its pixel footprint is too large to capture small features like houses, streets, or hedge rows. To address this, we present SEN4X, a hybrid super-resolution architecture that combines the advantages of single-image and multi-image techniques. It combines temporal oversampling from repeated Sentinel-2 acquisitions with a learned prior from high-resolution Pl\'eiades Neo data. In doing so, SEN4X upgrades Sentinel-2 imagery to 2.5 m ground sampling distance. We test the super-resolved images on urban land-cover classification in Hanoi, Vietnam. We find that they lead to a significant performance improvement over state-of-the-art super-resolution baselines.

Figures

Figures reproduced from arXiv: 2505.24799 by the authors.

Figure 1
Figure 1. Histogram of 𝐿1 -differences between LR images and down-sampled HR images before and after radiometric cross￾calibration (lower is better) [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Training, validation and test regions of the Hanoi dataset. A. Retnanto et al. Page 8 of 12 [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Architectures for combined SISR+MISR. Per default, we recommend the standard Sen4x, where multi-image fusion precedes single-image enhancement of the fused feature representation. A. Retnanto et al. Page 9 of 12 [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Comparison of SR results and LC segmentations for three exemplary scenes from Hanoi. Regions without ground truth labels are denoted by transparent masks. A. Retnanto et al. Page 10 of 12 [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

43 extracted references · 38 canonical work pages

  1. [1]

    L. Wald, T. Ranchin, M. Mangolini, Fusion of satellite images of differentspatialresolutions:Assessingthequalityofresultingimages, Photogrammetric Engineering and Remote Sensing 63 (6) (1997) 691–699

    work page 1997

  2. [2]

    B.Aiazzi,L.Alparone,S.Baronti,A.Garzelli,Context-drivenfusion of high spatial and spectral resolution images based on oversampled multiresolution analysis, IEEE Transactions on Geoscience and Re- mote Sensing 40 (10) (2002) 2300–2312

    work page 2002

  3. [3]

    J. Yang, X. Fu, Y. Hu, Y. Huang, X. Ding, J. Paisley, PanNet: A deep network architecture for pan-sharpening, in: IEEE International Conference on Computer Vision, 2017, pp. 5449–5457

    work page 2017

  4. [4]

    W. T. Freeman, T. R. Jones, E. C. Pasztor, Example-based super- resolution, IEEE Computer Graphics and Applications 22 (2) (2002) 56–65

    work page 2002

  5. [5]

    F. Li, X. Jia, D. Fraser, A. Lambert, Super resolution for remote sensingimagesbasedonauniversalhiddenmarkovtreemodel,IEEE TransactionsonGeoscienceandRemoteSensing48(3)(2009)1270– 1278

    work page 2009

  6. [6]

    Zhang, Y

    Y. Zhang, Y. Du, F. Ling, S. Fang, X. Li, Example-based super- resolution landcover mappingusing support vectorregression, IEEE JournalofSelectedTopicsinAppliedEarthObservationsandRemote Sensing 7 (4) (2014) 1271–1283

    work page 2014

  7. [7]

    Richard, I

    A. Richard, I. Cherabier, M. R. Oswald, V. Tsiminaki, M. Pollefeys, K. Schindler, Learned multi-view texture super-resolution, in: Inter- national Conference on 3D Vision, 2019, pp. 533–543

    work page 2019

  8. [8]

    Liebel, M

    L. Liebel, M. Körner, Single-image super resolution for multispec- tral remote sensing data using convolutional neural networks, The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences 41 (2016) 883–890

    work page 2016

Show all 43 references
  1. [9]

    Galar, R

    M. Galar, R. Sesma, C. Ayala, L. Albizua, C. Aranda, Super- resolution of Sentinel-2 images using convolutional neural networks and real ground truth data, Remote Sensing 12 (18) (2020) 2941

    2020

  2. [10]

    Rossi, V

    L. Rossi, V. Bernuzzi, T. Fontanini, M. Bertozzi, A. Prati, Swin2- MoSE: A new single image super-resolution model for remote sens- ing, arXiv preprint arXiv:2404.18924 (2024)

    2024 arXiv

  3. [11]

    Salgueiro Romero, J

    L. Salgueiro Romero, J. Marcello, V. Vilaplana, Super-resolution of Sentinel-2 imagery using generative adversarial networks, Remote Sensing 12 (15) (2020) 2424

    2020

  4. [12]

    Donike, C

    S. Donike, C. Aybar, L. Gómez-Chova, F. Kalaitzis, Trustworthy super-resolutionofmultispectralSentinel-2imagerywithlatentdiffu- sion,IEEEJournalofSelectedTopicsinAppliedEarthObservations and Remote Sensing (2025) 1–14

    2025

  5. [13]

    Y. Xiao, Q. Yuan, K. Jiang, J. He, X. Jin, L. Zhang, EDiffSR: An efficientdiffusionprobabilisticmodelforremotesensingimagesuper- resolution,IEEETransactionsonGeoscienceandRemoteSensing62 (2024) 1–14

    2024

  6. [14]

    Saharia, J

    C. Saharia, J. Ho, W. Chan, T. Salimans, D. J. Fleet, M. Norouzi, Imagesuper-resolutionviaiterativerefinement,IEEETransactionson Pattern Analysis and Machine Intelligence 45 (4) (2023) 4713–4726

    2023

  7. [15]

    N. L. Nguyen, J. Anger, A. Davy, P. Arias, G. Facciolo, L1BSR: Exploiting detector overlap for self-supervised single-image super- resolution of Sentinel-2 L1B imagery, in: IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 2013–2023

    2023

  8. [16]

    Deudon, A

    M. Deudon, A. Kalaitzis, I. Goytom, M. R. Arefin, Z. Lin, K. Sankaran, V. Michalski, S. E. Kahou, J. Cornebise, Y. Bengio, HighRes-net: Recursive fusion for multi-frame super-resolution of satellite imagery, arXiv preprint arXiv:2002.06460 (2020)

    2020 arXiv

  9. [17]

    M. T. Razzak, G. Mateo-García, G. Lecuyer, L. Gómez-Chova, Y. Gal, F. Kalaitzis, Multi-spectral multi-image super-resolution of Sentinel-2 with radiometric consistency losses and its effect on building delineation, ISPRS Journal of Photogrammetry and Remote Sensing 195 (2023) 1–13

    2023

  10. [18]

    Okabayashi, N

    A. Okabayashi, N. Audebert, S. Donike, C. Pelletier, Cross-sensor super-resolution of irregularly sampled Sentinel-2 time series, in: IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2024, pp. 502–511

    2024

  11. [19]

    Sainte Fare Garnot, L

    V. Sainte Fare Garnot, L. Landrieu, Lightweight temporal self- attention for classifying satellite images time series, in: Advanced Analytics and Learning on Temporal Data, Vol. 12588 of Lecture Notes in Computer Science, 2020, pp. 171–181

    2020

  12. [20]

    Cornebise, I

    J. Cornebise, I. Oršolić, F. Kalaitzis, Open high-resolution satel- lite imagery: The WorldStrat dataset – with application to super- resolution, arXiv preprint arXiv:2207.06418 (2022)

    2022 arXiv

  13. [21]

    Wolters, F

    P. Wolters, F. Bastani, A. Kembhavi, Zooming out on zooming in: Advancing super-resolution for remote sensing, arXiv preprint arXiv:2311.18082 (2023)

    2023 arXiv

  14. [22]

    Lanaras, J

    C. Lanaras, J. Bioucas-Dias, S. Galliani, E. Baltsavias, K. Schindler, Super-resolutionofSentinel-2images:Learningagloballyapplicable deepneural network,ISPRSJournal ofPhotogrammetryand Remote Sensing 146 (2018) 305–319

    2018

  15. [23]

    W. Dong, L. Mou, X. X. Zhu, Real-world remote sensing image super-resolutionviaapracticaldegradationmodelandakernel-aware network,ISPRSJournalofPhotogrammetryandRemoteSensing191 (2022) 155–169

    2022

  16. [24]

    H. Chen, X. He, L. Qing, Y. Wu, C. Ren, C. Zhu, Real- world single image super-resolution: A brief review, arXiv preprint arXiv:2103.02368 (2021)

    2021 arXiv

  17. [25]

    Z. Qiu, H. Shen, L. Yue, G. Zheng, Cross-sensor remote sensing imagerysuper-resolutionviaanedge-guidedattention-basednetwork, ISPRS Journal of Photogrammetry and Remote Sensing 199 (2023) 226–241

    2023

  18. [26]

    Z. Wang, A. Bovik, H. Sheikh, E. Simoncelli, Image quality assess- ment:fromerrorvisibilitytostructuralsimilarity,IEEETransactions on Image Processing 13 (4) (2004) 600–612

    2004

  19. [27]

    Zhang, P

    R. Zhang, P. Isola, A. A. Efros, E. Shechtman, O. Wang, The un- reasonable effectiveness of deep features as a perceptual metric, in: IEEE/CVFConferenceonComputerVisionandPatternRecognition, 2018

    2018

  20. [28]

    Aybar, D

    C. Aybar, D. Montero, S. Donike, F. Kalaitzis, L. Gómez-Chova, A comprehensive benchmark for optical remote sensing image super- resolution, IEEE Geoscience and Remote Sensing Letters 21 (2024) 1–5

    2024

  21. [29]

    European Space Agency, Sentinel-2 User Handbook,https: //sentinel.esa.int/documents/247904/685211/Sentinel-2_User_ Handbook(2015)

    2015

  22. [30]

    sentinel-hub.com

    Sentinel Hub, Cloud API for Satellite Imagery,https://www. sentinel-hub.com

  23. [31]

    J.Kuusk,Straylighteffectsinabove-waterremote-sensingreflectance fromhyperspectralradiometers,AppliedOptics55(15)(2016)3966– 3977

    2016

  24. [32]

    T. Bai, D. Li, K. Sun, Y. Chen, W. Li, Cloud detection for high- resolutionsatelliteimageryusingmachinelearningandmulti-feature fusion, Remote Sensing 8 (9) (2016) 715

    2016

  25. [33]

    H.-R.Hannula,J.Pulliainen,Spectralreflectancebehaviorofdifferent boreal snow types, Journal of Glaciology 65 (254) (2019) 926–939

    2019

  26. [34]

    pdf(2021)

    AirbusDefenceandSpace,PléiadesNeoUserGuide–EarlyVersion 3,https://wp-cdn.apollomapping.com/web_assets/user_uploads/2021/ 11/08103301/2021.10_PleiadesNeo_UserGuide-EarlyRelease_20211015. pdf(2021)

    2021

  27. [35]

    QGIS Development Team, QGIS Geographic Information System, QGIS Association,https://www.qgis.org

  28. [36]

    Sirko, S

    W. Sirko, S. Kashubin, M. Ritter, A. Annkah, Y. S. E. Bouchareb, Y. Dauphin, D. Keysers, M. Neumann, M. Cisse, J. Quinn, Continental-scale building detection from high resolution satellite imagery, arXiv preprint arXiv:2107.12283 (2021)

    2021 arXiv

  29. [37]

    M. V. Conde, U.-J. Choi, M. Burchi, R. Timofte, Swin2SR: SwinV2 transformer for compressed image super-resolution and restoration, in: European Conference on Computer Vision, 2022, pp. 669–687

    2022

  30. [38]

    D.P.Kingma,J.Ba,Adam:Amethodforstochasticoptimization,in: International Conference on Learning Representations, 2015

    2015

  31. [39]

    Bastani, P

    F. Bastani, P. Wolters, R. Gupta, J. Ferdinando, A. Kembhavi, Sat- laspretrain: A large-scale dataset for remote sensing image under- standing, arXiv preprint arXiv:2211.15660 (2023). A. Retnanto et al. Page 11 of 12

    2023 arXiv

  32. [40]

    Z. Liu, H. Hu, Y. Lin, Z. Yao, Z. Xie, Y. Wei, J. Ning, Y. Cao, Z. Zhang, L. Dong, F. Wei, B. Guo, Swin Transformer V2: Scaling up capacity and resolution, arXiv preprint arXiv:2111.09883 (2022)

    2022 arXiv

  33. [41]

    T.-Y. Lin, P. Dollar, R. Girshick, K. He, B. Hariharan, S. Belongie, Feature pyramid networks for object detection, in: IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, 2017

    2017

  34. [42]

    X. Wang, K. Yu, S. Wu, J. Gu, Y. Liu, C. Dong, Y. Qiao, C. Change Loy, ESRGAN: Enhanced super-resolution generative adversarial networks, in: European Conference on Computer Vision Workshops, 2018

    2018

  35. [43]

    Allen AI, Satlas super resolution,https://github.com/allenai/ satlas-super-resolution(2024). A. Retnanto et al. Page 12 of 12

    2024

Pith tools

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