Pith. sign in

REVIEW 3 major objections 7 minor 46 references

SAR Strikes Back: A New Hope for RSVQA

T0 review · 3 major / 7 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read By turning SAR pixels into a short text list of land-cover classes, a two-stage pipeline answers remote-sensing questions about radar imagery almost 10 percentage points better than an end-to-end model, and late fusion with optical…

desk verdict The split-leakage worry is overstated—the paper says train/test are geographically separated—but the missing split spec, no error bars, and no data release still make this a major-revision rather than a clean accept. read the letter →

arxiv 2501.08131 v2 pith:SLDKWOXZ submitted 2025-01-14 cs.CV

Lucrezia Tosato , Flora Weissgerber , Laurent Wendling , Sylvain Lobry This is my paper · ORCID
classification cs.CV
keywords DeepLearningRemoteSensingVisualQuestionAnsweringMulti-ModalityNaturalLanguageProcessingSyntheticApertureRadardatafusionlandcoverclassification
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

Remote-sensing visual question answering (RSVQA) has mostly been an optical-image task; this paper argues that synthetic aperture radar (SAR) can be brought in productively. It introduces RSVQAxBEN-MM, a benchmark that pairs Sentinel-1 radar patches (VV, VH, and a normalized VV/VH ratio) with the existing Sentinel-2 optical questions of RSVQAxBEN, and it studies two pipelines on it. The central claim is that a two-stage pipeline—first classifying the image into a text list of land-cover classes, then letting a language model answer from that list—works far better with SAR than an end-to-end model, gaining nearly 10 percentage points in overall accuracy. The paper further claims that fusing SAR and optical at the decision level, by combining the two classifiers' per-class scores, gives the best results: 75.49% overall accuracy on the VQA test set, with the late-fusion classifier at 75.00% F1-micro. If true, this is a practical route to all-weather, interpretable land-cover question answering, with SAR carrying information optical images miss, especially over water.

What carries the argument

The load-bearing mechanism is the text bottleneck: the Prompt-RSVQA context generated by thresholding multi-label classification scores. By converting image content into a list of class names, it aligns radar-derived information with the language model's pretrained word space, which is why the paper sees SAR-only gains over end-to-end and why late fusion of class-score vectors outperforms feature-level fusion.

What would settle it

Retrain the BEN-MM-61 classifier on a strict geographic split (western-Europe patches for training, eastern-Europe for testing, mirroring RSVQAxBEN), then rerun Prompt-RSVQA with SAR, optical, and late fusion on the VQA test set. If the ~10-point SAR-only gain over end-to-end and the late-fusion lead shrink materially, the mismatch between the classifier's training split and the VQA test split is the cause.

Watch

Extended reading notes

Core claim

On its own terms, the paper's claim is that SAR imagery can be brought into RSVQA not by feeding radar pixels through one monolithic network, but by a two-stage translation: a multi-label land-cover classifier turns the SAR patch into a short text list (for example 'Forest, Water bodies'), and a language model answers the question from that list plus the question text. The paper constructs RSVQAxBEN-MM, pairing each Sentinel-2 RGB patch and its existing RSVQAxBEN question-answer triplets with the matching Sentinel-1 dual-polarization patch including a normalized VV/VH ratio channel. In experiments, this Prompt-RSVQA pipeline with SAR alone reaches 71.78% overall accuracy, nearly 10 percentage points above the best end-to-end SAR result, and the best fusion—late fusion, which concatenates the separate optical and SAR class-score vectors through a small MLP—reaches 75.49% VQA accuracy and 75.00% F1-micro on the classification step. The paper also argues SAR contributes specifically to water-body classes, and that the two-stage design makes errors inspectable through the intermediate class list.

Load-bearing premise

The paper never states which train/test split is used to train the image classifier behind Prompt-RSVQA, and the original BigEarthNet split is random while the VQA test split is geographic; if the classifier saw the VQA test patches during training, the reported gains are inflated.

Editorial extensions

If this is right

  • SAR-only two-stage RSVQA reaches 71.78% overall accuracy, roughly 10 points over end-to-end SAR, so SAR can support land-cover question answering without optical imagery.
  • Decision-level fusion of independently trained SAR and optical classifiers yields the best VQA accuracy (75.49%) and best classification F1-micro (75.00%) on the proposed dataset.
  • Fusion gains are class-dependent: late and halfway fusion improve water-body predictions, while wetlands remain poorly predicted, so SAR's complementary value is not uniform across land-cover types.
  • The two-stage pipeline's intermediate class text makes answer errors traceable to misclassified land-cover classes, supporting error analysis.
  • Because the proposed dataset uses an east-west geographic split, the reported numbers are meant to measure geographic generalization rather than in-distribution memorization.

Reading between the lines

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

  • Because the paper never states the train/test split used to fit the BEN-MM-61 classifier, the cleanest way to stress-test the headline is to retrain that classifier on the same geographic split as RSVQAxBEN and watch whether the ~10-point SAR gain survives; this is the immediate next experiment.
  • The text-bottleneck design can express only what the fixed land-cover vocabulary names; spatial relations, counts, and object locations would require object-level or region-level context generation, which the current dataset does not exercise.
  • The fusion ranking is probably class-frequency-dependent: late fusion wins on the frequent and water classes that dominate questions, so a more balanced question distribution could reverse the ranking—the paper itself acknowledges the winner is dataset-dependent.
  • A direct transfer test to other dual-polarization SAR sensors, or to cloud-covered optical scenes, would show whether the VV/VH/ratio recipe generalizes beyond Sentinel-1.
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. The paper introduces RSVQAxBEN-MM, a dataset pairing Sentinel-1 SAR images with the existing RSVQAxBEN question/answer set, and compares an End-to-End RSVQA pipeline against a two-stage Prompt-RSVQA pipeline that converts a visual classifier's output into textual context for a language model. Experiments cover optical-only, SAR-only, and three fusion schemes (early, halfway, late). The headline claims are that Prompt-RSVQA improves overall accuracy by nearly 10% over End-to-End on SAR-only inputs and that late fusion yields the best VQA accuracy, 75.49% overall with F1-micro 75.00%.

Significance. If the reported results are trustworthy, this is a useful contribution: RSVQAxBEN-MM is a sensible extension of an established benchmark, the paper systematically compares SAR and optical modalities for RSVQA, and the analysis of per-class fusion behavior (e.g., SAR helping water-body classes) is informative. The paper also includes a clear discussion of dataset biases and releases detailed classification and VQA tables. The central quantitative claims, however, rest on an experimental protocol that is incompletely specified (the classifier split is not documented) and on single-seed runs with no error bars, so the magnitude of the reported improvements is not yet established.

major comments (3)
  1. [§VI-A, §IV-A, §IV-B] The split used to train the BEN-MM-61 image classifier is never specified. Section IV-A states that the original BEN benchmark uses a random 60/20/20 patch split, while Section IV-B defines RSVQAxBEN's geographic split with the easternmost 23% of patches as the test set. Section VI-A only says 'We first train image encoders on the BEN-MM-61 classification task' without saying which split is used. If the classifier follows the standard random BEN split, then a substantial fraction of the RSVQAxBEN-MM test patches (those in the eastern region) will have been seen during classifier training. Since Prompt-RSVQA converts the classifier's predicted class list into the textual context used to answer the question, this creates a direct label-leakage path for the test set, which would inflate the ~10% accuracy gap in Table V[d] versus Table IV[j] and the fusion gains. The authors must state the split used; if it was random, they should retrain the classifier on the RSVQAxBEN-MM training split (or otherwise ensure no overlap) and rerun the Prompt-RSVQA and fusion experiments.
  2. [§VI-B/C, Tables IV–V] All experiments are reported as single runs with no standard deviations or significance testing. This is particularly problematic for the claimed best result: the late-fusion overall accuracy of 75.49% (Table V[g]) differs from the optical-only result of 75.40% (Table V[b]) by only 0.09 percentage points, which is well within typical run-to-run variability for deep models. Similarly, the SAR-only versus optical-only differences in Table IV are small. Without multiple seeds (at least 3, preferably 5) with mean and standard deviation, or a statistical test, the superiority of late fusion and the value of SAR as a complementary modality are not convincingly demonstrated.
  3. [§VII and Figure 4] The claim that 'SAR proves especially beneficial for questions related to specific land cover types, such as water areas' is supported primarily by classification F1 variations in Figure 4 and by qualitative examples in Figure 5, but there is no quantitative VQA accuracy breakdown by class or by question type for SAR versus optical versus fusion. The paper should report per-class or per-question-type VQA accuracy (e.g., for questions mentioning water bodies or wetlands) to substantiate this claim, since the overall VQA accuracy differences are small and possibly within noise.
minor comments (7)
  1. [Abstract/Introduction] The abstract states that 'no study has compared SAR and optical imagery in RSVQA,' but reference [8] (by the same authors) appears to be exactly such a comparison; the distinction (e.g., very-high-resolution SAR vs. Sentinel-1, or question/answer design) should be made explicit to avoid overclaiming novelty.
  2. [§IV-D heading] The heading 'D. Dataset evaluation)' contains a stray closing parenthesis; it should read 'D. Dataset evaluation'.
  3. [Table III and §VI-A] The table caption says 'Classification results on BEN-MM' while the text consistently refers to 'BEN-MM-61'; please unify the naming to avoid confusion about the label set.
  4. [§VI-B] The sentence 'The results of the 10 experiments are presented in Table IV' is fine, but the description of which rows are unfrozen is only given for ResNet-50 initialized on ImageNet; clarify explicitly which rows (e.g., e and j) correspond to fine-tuned networks.
  5. [Tables IV and V] The LBscore column is reported but the metric is not defined in Section V; add a definition (or a reference to [41]) in the performance-evaluation section.
  6. [Figure 6] The confusion matrix is extremely dense and difficult to read, especially the bottom-right zoom; consider splitting into separate figures or increasing resolution.
  7. [§VII-A] The sentence beginning 'In Table III[a,b] we can verify that optical results lose 1.6%' is grammatically incomplete; it should specify what the 1.6% is relative to (presumably ResNet-152 vs. ResNet-50) and complete the thought.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found; the claimed VQA gains are empirical outputs of a trained two-stage pipeline, with the unspecified BEN-MM-61 split being a leakage risk rather than a derivational circularity.

full rationale

This is an empirical systems paper rather than a derivation, and its claimed results are not equivalent to their inputs by construction. Prompt-RSVQA's VQA output is a trained language-model function of the textual class context produced by the BEN-MM-61 classifier; the fact that the context is derived from CLC labels, which also define the VQA ground truth, is the intended design of the two-stage prompt pipeline, not a hidden reduction. The paper explicitly separates classification metrics (F1, match ratio, Hamming distance) from VQA accuracy and observes that classification gains do not translate directly into VQA gains, so VQA accuracy is not being presented as a renamed classifier score. The self-citations ([7], [8], [41]) are used for background, method inspiration, and interpretative support; none is load-bearing in the sense of forcing the reported numbers. The main validity concern is experimental: the split used to train the BEN-MM-61 image encoders is not stated (Section VI-A), and because BEN uses a random split while RSVQAxBEN-MM uses a geographic split, there is a possible overlap between the classifier training set and the VQA test set. That is a data-leakage or reproducibility risk, not a circularity, and it cannot be exhibited as a reduction from the paper's own text because the classifier split is unspecified. Under the rule that circularity requires a quotable reduction, no circular step is established.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The paper's contributions are empirical and rest on transferable pre-trained features, the validity of CLC-derived labels for SAR imagery, and clean dataset splits. The most serious unaccounted premise is the split overlap between classifier pre-training and the VQA test set.

free parameters (1)
  • Classification threshold for context binarization = not reported
    In Prompt-RSVQA, the predicted class scores fc are thresholded to produce the text context c (Section III-B). The threshold value is never stated, yet it directly determines the context and therefore the VQA answer.
assumptions (3)
  • domain assumption RSVQAxBEN question/answer pairs, generated from CLC labels of the optical Sentinel-2 patches, are valid ground truth for the corresponding Sentinel-1 SAR patches.
    The dataset in Section IV-C pairs SAR images with RSVQAxBEN triplets. The answers describe land cover classes originally derived from CLC via optical imagery; the paper assumes these labels also characterize the SAR patch despite different acquisition time, geometry, and backscatter.
  • domain assumption The classification pre-training split for BEN-MM-61 is disjoint from the RSVQA test set.
    Section VI-A trains encoders on BEN-MM-61 classification, while Section IV-B defines the RSVQA split by geography. The paper never states whether the classification split respects the RSVQA split; the original BEN benchmark uses a random split, so test patches may have been seen during classifier training.
  • domain assumption ImageNet pretraining transfers to non-optical SAR inputs once the input channels are adjusted.
    Used for SAR-only and early fusion ResNets (Section VI-A). SAR backscatter has very different statistics from natural RGB images; no evidence is provided that ImageNet features remain useful for VV/VH/ratio inputs.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SAR Strikes Back: A New Hope for RSVQA." pith.science (2026). https://pith.science/paper/SLDKWOXZ

@misc{pith2026250108131,
  author       = {Pith},
  title        = {Pith review of: SAR Strikes Back: A New Hope for RSVQA},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SLDKWOXZ}},
  note         = {Machine review of arXiv:2501.08131}
}
read the original abstract

Remote Sensing Visual Question Answering (RSVQA) is a task that extracts information from satellite images to answer questions in natural language, aiding image interpretation. While several methods exist for optical images with varying spectral bands and resolutions, only recently have high-resolution Synthetic Aperture Radar (SAR) images been explored. SAR's ability to operate in all weather conditions and capture electromagnetic features makes it a promising modality, yet no study has compared SAR and optical imagery in RSVQA or proposed effective fusion strategies. This work investigates how to integrate SAR data into RSVQA and how to best combine it with optical images. We present a dataset that enables SAR-based RSVQA and explore two pipelines for the task. The first is an end-to-end model, while the second is a two-stage framework: SAR information is first extracted and translated into text, which is then processed by a language model to produce the final answer. Our results show that the two-stage model performs better, improving accuracy by nearly 10% over the end-to-end approach. We also evaluate fusion strategies for combining SAR and optical data. A decision-level fusion yields the best results, with an F1-micro score of 75.00%, F1-average of 81.21%, and overall accuracy of 75.49% on the proposed dataset. SAR proves especially beneficial for questions related to specific land cover types, such as water areas, demonstrating its value as a complementary modality to optical imagery.

Figures

Figures reproduced from arXiv: 2501.08131 by the authors.

Figure 1
Figure 1. The two pipelines proposed in this work. In both pipelines, we extract the visual information (an abstract visual feature [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The different visual models studied in this work. (a) and (b) are mono-modalities models. In both cases, we use a CNN [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Workflow from the initial BEN dataset containing the Sentinel-2 optical images and their CLC classes, to BEN-MM-61 [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Percental variation in the behaviour of each L1 class based on the F1-average score, comparing results from the three [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: Visual results from the test set. The two modalities, the prediction of OPTICAL and SAR image only and the late [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: Confusion matrices of the Prompt-RSVQA model with a ResNet-50 backbone and a late fusion mechanism. On the [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

46 extracted references · 42 canonical work pages

  1. [1]

    Improved marine debris detection in satellite imagery with automatic refinement of coarse hand annotations,

    M. C. Russwurm, D. G ¨ul, and D. Tuia, “Improved marine debris detection in satellite imagery with automatic refinement of coarse hand annotations,” in ICLR Workshops, 2023

    work page 2023

  2. [2]

    Half a percent of labels is enough: Efficient animal detection in UA V imagery using deep CNNs and active learning,

    B. Kellenberger, D. Marcos, S. Lobry, and D. Tuia, “Half a percent of labels is enough: Efficient animal detection in UA V imagery using deep CNNs and active learning,” TGRS, vol. 57, no. 12, pp. pp. 9524–9533, 2019

    work page 2019

  3. [3]

    Domain adaptation for mapping lczs in sub-saharan africa with remote sensing: A comprehensive approach to health data analysis,

    B. Rousse, S. Lobry, G. Duth ´e, V . Golaz, and L. Wendling, “Domain adaptation for mapping lczs in sub-saharan africa with remote sensing: A comprehensive approach to health data analysis,” JSTARS, 2024

    work page 2024

  4. [4]

    Burning skies: The hid- den toll of big oil’s toxic flames https://eiforum.org/ burning-skies-the-hidden-of-big-oils-toxic-flames/,

    EIF Investigations, 2024., “Burning skies: The hid- den toll of big oil’s toxic flames https://eiforum.org/ burning-skies-the-hidden-of-big-oils-toxic-flames/,” (Accessed 03/10/2024). 1https://land.copernicus.eu/content/corine-land-cover-nomenclature-guidelines/ html/

    work page 2024

  5. [5]

    RSVQA: Visual question answering for remote sensing data,

    S. Lobry, D. Marcos, J. Murray, and D. Tuia, “RSVQA: Visual question answering for remote sensing data,” TGRS, vol. 58, no. 12, pp. pp. 8555–8566, 2020

    work page 2020

  6. [6]

    VQA: Visual question answering,

    S. Antol, A. Agrawal, J. Lu, M. Mitchell, D. Batra, C. L. Zitnick, and D. Parikh, “VQA: Visual question answering,” in ICCV. IEEE, 2015, pp. 2425–2433

    work page 2015

  7. [7]

    Prompt- RSVQA: Prompting visual context to a language model for remote sensing visual question answering,

    C. Chappuis, V . Zermatten, S. Lobry, B. Le Saux, and D. Tuia, “Prompt- RSVQA: Prompting visual context to a language model for remote sensing visual question answering,” in CVPR, 2022, pp. 1372–1381

    work page 2022

  8. [8]

    Can SAR improve RSVQA performance?

    L. Tosato, S. Lobry, F. Weissgerber, and L. Wendling, “Can SAR improve RSVQA performance?” in EUSAR. VDE, 2024, pp. 1287– 1292

    work page 2024

Show all 46 references
  1. [9]

    Deep learning meets SAR: Concepts, models, pitfalls, and perspectives,

    X. X. Zhu, S. Montazeri, M. Ali, Y . Hua, Y . Wang, L. Mou, Y . Shi, F. Xu, and R. Bamler, “Deep learning meets SAR: Concepts, models, pitfalls, and perspectives,” IEEE GRS, vol. 9, no. 4, pp. 143–172, 2021

    2021

  2. [10]

    A review of the application of optical and radar remote sensing data fusion to land use mapping and monitoring,

    N. Joshi, M. Baumann, A. Ehammer, R. Fensholt, K. Grogan, P. Hostert, M. R. Jepsen, T. Kuemmerle, P. Meyfroidt, E. T. Mitchard et al. , “A review of the application of optical and radar remote sensing data fusion to land use mapping and monitoring,” Remote Sensing, vol. 8, no....

    2016

  3. [11]

    Language transformers for remote sensing visual question answering,

    C. Chappuis, V . Mendez, E. Walt, S. Lobry, B. Le Saux, and D. Tuia, “Language transformers for remote sensing visual question answering,” in IGARSS. IEEE, 2022, pp. 4855–4858

    2022

  4. [12]

    How to find a good image-text embedding for remote sensing visual question answering?

    C. Chappuis, S. Lobry, B. A. Kellenberger, B. Le Saux, and D. Tuia, “How to find a good image-text embedding for remote sensing visual question answering?” in ECML Workshops, 2021

    2021

  5. [13]

    Segmentation-guided attention for visual question answering from remote sensing images,

    L. Tosato, H. Boussaid, F. Weissgerber, C. Kurtz, L. Wendling, and S. Lobry, “Segmentation-guided attention for visual question answering from remote sensing images,” in IGARSS. IEEE, 2024, pp. 2750–2754

    2024

  6. [14]

    Multilingual augmentation for robust visual question answering in remote sensing images,

    Z. Yuan, L. Mou, and X. X. Zhu, “Multilingual augmentation for robust visual question answering in remote sensing images,” in JURSE. IEEE, 2023, pp. 1–4

    2023

  7. [15]

    Multi-task prompt-RSVQA to explicitly count objects on aerial images

    C. Chappuis, C. Sertic, N. Santacroce, J. C. Navarro, S. Lobry, B. Le Saux, and D. Tuia, “Multi-task prompt-RSVQA to explicitly count objects on aerial images.” in BMVC Workshop, 2023. IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERV ATIONS AND REMOTE SENSING, VOL. XX, ...

    2023

  8. [16]

    Remote sensing chatgpt: Solving remote sensing tasks with chatgpt and visual models,

    H. Guo, X. Su, C. Wu, B. Du, L. Zhang, and D. Li, “Remote sensing chatgpt: Solving remote sensing tasks with chatgpt and visual models,” arXiv preprint arXiv:2401.09083 , 2024

    2024 arXiv

  9. [17]

    Visual Question Answering for Wishart H-Alpha Classification of Polarimetric SAR Images,

    H. Aghababaei and A. Stein, “Visual Question Answering for Wishart H-Alpha Classification of Polarimetric SAR Images,” inIGARSS. IEEE, 2024, pp. 11 231–11 234

    2024

  10. [18]

    A visual question answering method for SAR ship: Breaking the requirement for multimodal dataset construction and model fine-tuning,

    F. Wang, C. Chen, H. Chen, Y . Chang, and W. Zeng, “A visual question answering method for SAR ship: Breaking the requirement for multimodal dataset construction and model fine-tuning,” arXiv preprint arXiv:2411.01445, 2024

    2024 arXiv

  11. [19]

    Text-guided coarse-to-fine fusion network for robust remote sensing visual question answering,

    Z. Zhao, C. Zhou, Y . Zhang, C. Li, X. Ma, and J. Tang, “Text-guided coarse-to-fine fusion network for robust remote sensing visual question answering,” arXiv preprint arXiv:2411.15770 , 2024

    2024 arXiv

  12. [20]

    SAR target recognition based on deep learning,

    S. Chen and H. Wang, “SAR target recognition based on deep learning,” in DSAA. IEEE, 2014, pp. 541–547

    2014

  13. [21]

    SAR image despeckling through convolutional neural networks,

    G. Chierchia, D. Cozzolino, G. Poggi, and L. Verdoliva, “SAR image despeckling through convolutional neural networks,” in IGARSS. IEEE, 2017, pp. 5438–5441

    2017

  14. [22]

    Towards global volcano moni- toring using multisensor sentinel missions and artificial intelligence: The mounts monitoring system,

    S. Valade, A. Ley, F. Massimetti, O. D’Hondt, M. Laiolo, D. Coppola, D. Loibl, O. Hellwich, and T. R. Walter, “Towards global volcano moni- toring using multisensor sentinel missions and artificial intelligence: The mounts monitoring system,” Remote Sensing , vol. 11, no. 13, ...

    2019

  15. [23]

    Automatic sim- ulation of SAR images: comparing a deep-learning based method to a hybrid method,

    N. Letheule, F. Weissgerber, S. Lobry, and E. Colin, “Automatic sim- ulation of SAR images: comparing a deep-learning based method to a hybrid method,” in IGARSS. IEEE, 2023, pp. 4958–4961

    2023

  16. [24]

    Exploring data and models in SAR ship image captioning,

    K. Zhao and W. Xiong, “Exploring data and models in SAR ship image captioning,” IEEE Access, vol. 10, pp. pp. 91 150–91 159, 2022

    2022

  17. [25]

    An im- proved SAR ship classification method using text-to-image generation- based data augmentation and squeeze and excitation,

    L. Wang, Y . Qi, P. T. Mathiopoulos, C. Zhao, and S. Mazhar, “An im- proved SAR ship classification method using text-to-image generation- based data augmentation and squeeze and excitation,” Remote Sensing, vol. 16, no. 7, p. pp. 1299, 2024

    2024

  18. [26]

    Learning capsules for SAR target recognition,

    Y . Guo, Z. Pan, M. Wang, J. Wang, and W. Yang, “Learning capsules for SAR target recognition,” JSTARS, vol. 13, pp. pp. 4663–4673, 2020

    2020

  19. [27]

    A deep learning framework for matching of SAR and optical imagery,

    L. H. Hughes, D. Marcos, S. Lobry, D. Tuia, and M. Schmitt, “A deep learning framework for matching of SAR and optical imagery,” ISPRS, vol. 169, pp. pp. 166–179, 2020

    2020

  20. [28]

    A theoretical and practical survey of image fusion methods for multispectral pansharpening,

    C. S. Yilmaz, V . Yilmaz, and O. Gungor, “A theoretical and practical survey of image fusion methods for multispectral pansharpening,” In- formation Fusion, vol. 79, pp. pp. 1–43, 2022

    2022

  21. [29]

    Urban mapping using coarse SAR and optical data: Outcome of the 2007 GRSS data fusion contest,

    F. Pacifici, F. Del Frate, W. J. Emery, P. Gamba, and J. Chanussot, “Urban mapping using coarse SAR and optical data: Outcome of the 2007 GRSS data fusion contest,” GRSL, vol. 5, no. 3, pp. pp. 331–335, 2008

    2007

  22. [30]

    Pedestrian detection based on deep fusion network using feature correlation,

    Y . Lee, T. D. Bui, and J. Shin, “Pedestrian detection based on deep fusion network using feature correlation,” in APSIPA ASC. IEEE, 2018, pp. 694–699

    2018

  23. [31]

    Decision fusion for the classification of hyperspectral data: Outcome of the 2008 GRS-S data fusion contest,

    G. Licciardi, F. Pacifici, D. Tuia, S. Prasad, T. West, F. Giacco, C. Thiel, J. Inglada, E. Christophe, J. Chanussot et al. , “Decision fusion for the classification of hyperspectral data: Outcome of the 2008 GRS-S data fusion contest,” TGRS, vol. 47, no. 11, pp. pp. 3857–3865, 2009

    2008

  24. [32]

    Early vs late fusion in multimodal convolutional neural networks,

    K. Gadzicki, R. Khamsehashari, and C. Zetzsche, “Early vs late fusion in multimodal convolutional neural networks,” inFUSION. IEEE, 2020, pp. 1–6

    2020

  25. [33]

    Multispectral deep neural networks for pedestrian detection,

    J. Liu, S. Zhang, S. Wang, and D. N. Metaxas, “Multispectral deep neural networks for pedestrian detection,” in BMVC, 2016

    2016

  26. [34]

    Multispectral pedestrian detection using deep fusion convolutional neural networks

    J. Wagner, V . Fischer, M. Herman, S. Behnke et al. , “Multispectral pedestrian detection using deep fusion convolutional neural networks.” in ESANN, vol. 587, 2016, pp. 509–514

    2016

  27. [35]

    Challenges and opportunities of multimodality and data fusion in remote sensing,

    M. Dalla Mura, S. Prasad, F. Pacifici, P. Gamba, and J. Chanussot, “Challenges and opportunities of multimodality and data fusion in remote sensing,” in EUSIPCO. IEEE, 2014, pp. 106–110

    2014

  28. [36]

    Earthgpt: A universal multi-modal large language model for multi-sensor image comprehension in remote sensing domain,

    W. Zhang, M. Cai, T. Zhang, Y . Zhuang, and X. Mao, “Earthgpt: A universal multi-modal large language model for multi-sensor image comprehension in remote sensing domain,” IEEE TGRS, 2024

    2024

  29. [37]

    BigEarthNet: A large-scale benchmark archive for remote sensing image understanding [dataset],

    G. Sumbul, M. Charfuelan, B. Demir, and V . Markl, “BigEarthNet: A large-scale benchmark archive for remote sensing image understanding [dataset],” in IGARSS. IEEE, 2019, pp. 5901–5904

    2019

  30. [38]

    BigEarthNet-MM: A large- scale, multimodal, multilabel benchmark archive for remote sensing image classification and retrieval [dataset],

    G. Sumbul, A. De Wall, T. Kreuziger, F. Marcelino, H. Costa, P. Benev- ides, M. Caetano, B. Demir, and V . Markl, “BigEarthNet-MM: A large- scale, multimodal, multilabel benchmark archive for remote sensing image classification and retrieval [dataset],” GRSS, vol. 9, no. 3, pp...

    2021

  31. [39]

    Quantitative comparison of classification capability: Fully polarimetric versus dual and single- polarization SAR,

    J.-S. Lee, M. R. Grunes, and E. Pottier, “Quantitative comparison of classification capability: Fully polarimetric versus dual and single- polarization SAR,” TGRS, vol. 39, no. 11, pp. pp. 2343–2351, 2001

    2001

  32. [40]

    Integration of multitemporal/polarization c- band SAR data sets for land-cover classification,

    N.-W. Park and K.-H. Chi, “Integration of multitemporal/polarization c- band SAR data sets for land-cover classification,” IJRS, vol. 29, no. 16, pp. pp. 4667–4688, 2008

    2008

  33. [41]

    The curse of language biases in remote sensing VQA: the role of spatial attributes, language diversity, and the need for clear evaluation,

    C. Chappuis, E. Walt, V . Mendez, S. Lobry, B. L. Saux, and D. Tuia, “The curse of language biases in remote sensing VQA: the role of spatial attributes, language diversity, and the need for clear evaluation,” arXiv preprint arXiv:2311.16782, 2023

    2023 arXiv

  34. [42]

    Space: Structured compression and sharing of representational space for continual learning,

    G. Saha, I. Garg, A. Ankit, and K. Roy, “Space: Structured compression and sharing of representational space for continual learning,” IEEE Access, vol. 9, pp. pp. 150 480–150 494, 2021

    2021

  35. [43]

    Wetland monitoring using sar data: A meta- analysis and comprehensive review,

    S. Adeli, B. Salehi, M. Mahdianpari, L. J. Quackenbush, B. Brisco, H. Tamiminia, and S. Shaw, “Wetland monitoring using sar data: A meta- analysis and comprehensive review,” Remote Sensing , vol. 12, no. 14, p. 2190, 2020

    2020

  36. [44]

    Integrating sar and optical remote sensing for conservation-targeted wetlands mapping,

    H. Sahour, K. M. Kemink, and J. O’connell, “Integrating sar and optical remote sensing for conservation-targeted wetlands mapping,” Remote Sensing, vol. 14, no. 1, p. 159, 2021

    2021

  37. [45]

    Visualbert: A simple and performant baseline for vision and language,

    L. H. Li, M. Yatskar, D. Yin, C.-J. Hsieh, and K.-W. Chang, “Visualbert: A simple and performant baseline for vision and language,” arXiv preprint arXiv:1908.03557, 2019

    1908 arXiv

  38. [46]

    Multi-modal fusion transformer for visual question answering in remote sensing,

    T. Siebert, K. N. Clasen, M. Ravanbakhsh, and B. Demir, “Multi-modal fusion transformer for visual question answering in remote sensing,” in Image and Signal Processing for Remote Sensing XXVIII , vol. 12267. SPIE, 2022, pp. 162–170. IEEE JOURNAL OF SELECTED TOPICS IN APPLIED ...

    2022

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.