Pith. sign in

REVIEW 4 major objections 6 minor 33 references

A Novel Lightweight Transformer with Edge-Aware Fusion for Remote Sensing Image Captioning

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A lightweight transformer that fuses edge maps into its input reports the best captioning scores on three remote-sensing benchmarks.

desk verdict A clearly written engineering combination with an uncontrolled benchmark: the claimed SOTA margins rest on comparing corrected-caption results against original-caption baselines. read the letter →

arxiv 2506.09429 v1 pith:LIS4TZCH submitted 2025-06-11 cs.CV

Swadhin Das , Divyansh Mundra , Priyanshu Dayal , Raksha Sharma This is my paper · ORCID
classification cs.CV
keywords remotesensingimagecaptioninglightweighttransformerknowledgedistillationedgedetectionfusionConvNeXtDistilGPT2CIDEr
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 tries to show that a deliberately shrunk transformer model can match or beat much larger captioning systems if it is trained with knowledge distillation and fed edge-enhanced images alongside the originals. It claims that this combination, called KD-EBF, improves caption quality on the SYDNEY, UCM, and RSICD datasets across BLEU-1 to BLEU-4, METEOR, ROUGE-L, and CIDEr. The practical payoff is that accurate remote-sensing image captioning could run with far fewer parameters, making it usable in resource-constrained settings.

What carries the argument

The core mechanism is a Parameter Reduction Factor (PRF) that proportionally shrinks channel counts, normalization dimensions, and fully connected layers of the ConvNeXt encoder, with weights resized by interpolation. Knowledge distillation transfers the full teacher's behavior into this smaller student. An edge-aware fusion step extracts an edge map via Canny, Sobel, or Laplacian, concatenates it to the original RGB image, and feeds six-channel input through a modified first convolutional layer, so structural boundary cues enter the network without a second processing stream.

What would settle it

Run the same baseline models on the corrected SYDNEY, UCM, and RSICD splits and captions from [7]. If their BLEU and CIDEr numbers match the values copied into Tables XIII-XV, the claimed gaps hold; if their numbers change substantially, the reported improvement is largely an artifact of dataset revision rather than architecture.

Watch

Extended reading notes

Core claim

The paper's central claim is that KD-EBF, an encoder-decoder model using a dimension-reduced ConvNeXt transformer encoder, a DistilGPT2 decoder, knowledge distillation from a full teacher, and an edge-detection-based input fusion, outperforms all listed baselines on all seven metrics on all three datasets. The authors further find that distilled lightweight versions with parameter reduction factors of 2 or 3 match or exceed the original model, and that adding edge information consistently improves results, with Canny best on SYDNEY and Laplacian best on UCM and RSICD.

Load-bearing premise

The headline comparisons assume the published baseline scores, produced on the original captions, are directly comparable to the proposed model trained and evaluated on the corrected datasets of [7].

Editorial extensions

If this is right

  • The six-channel input design lets a captioning model exploit edge structure without doubling the encoder, preserving the lightweight goal.
  • Distillation recovers most or all of the accuracy lost when the encoder is shrunk, so parameter reduction does not have to cost caption quality.
  • Edge fusion gives consistent metric gains on all three datasets, suggesting boundary information is a broadly useful complement to semantic features in remote-sensing scenes.
  • KD-EBF's reported numbers on SYDNEY, UCM, and RSICD would be the new best known results for these benchmarks if the comparisons hold.
  • The resulting model has a much smaller decoder footprint than GPT-2 or RoBERTa, easing deployment on limited hardware.

Reading between the lines

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

  • A direct implication the paper leaves implicit is that the same edge-fusion trick could be dropped into other lightweight image-to-text models, not just this specific architecture, and likely give them a similar boost.
  • Because the paper trains and evaluates on the corrected captions of [7] while the baseline tables copy numbers from papers that used the original captions, re-running the baselines on the corrected datasets would reveal how much of the reported gap comes from data revision rather than from the model design.
  • The PRF-based shrinkage recipe, with interpolation-based weight resizing, could be tested as a general cheap way to slim other CNN-transformer encoders in vision-language tasks.
  • The human evaluation uses a single annotator; repeating the related/partially-related/unrelated rating with several annotators would test whether the perceived quality advantage is stable.
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. This paper proposes a lightweight transformer-based remote sensing image captioning model (KD-EBF) that combines a ConvNeXt encoder with reduced channel dimensions through a Parameter Reduction Factor (PRF), a DistilGPT2 decoder, knowledge distillation from a larger teacher model, and an edge-enhanced input fusion strategy that concatenates an edge map (Canny, Sobel, or Laplacian) with the original image as a six-channel input. The paper evaluates on SYDNEY, UCM, and RSICD datasets and reports superior results over several baselines including the authors' previous TextGCN and Conv-GPT models, with extensive ablations over PRF values, edge detectors, and decoder types.

Significance. If the reported results are valid, the paper would make a useful empirical contribution by demonstrating that a substantially smaller transformer encoder with a distilled GPT-2 decoder and edge-augmented inputs can match or exceed the caption quality of much larger models, which is relevant for practical deployment. The systematic ablations of PRF, edge detectors, and decoders are valuable, and the paper includes a human evaluation. However, the main experimental claim of state-of-the-art performance is currently confounded by the use of corrected datasets for the proposed model while citing baseline numbers from papers that used the original datasets, an unspecified knowledge distillation procedure, and test-set-based hyperparameter selection without statistical significance. These issues must be addressed before the contribution can be accepted.

major comments (4)
  1. [Section IV.B and Tables XIII-XV] The paper states in Section IV.B that "we have used the modified version of these datasets [7]" for all experiments, yet the baseline scores in Tables XIII-XV are taken from published papers that evaluated on the original, uncorrected datasets. Since all seven metrics (BLEU-1 through BLEU-4, METEOR, ROUGE-L, CIDEr) measure n-gram overlap with reference captions, altering the reference text changes the metric values even for identical model outputs. The paper must re-evaluate all baselines on the corrected datasets (or use the original datasets for all methods) to provide a fair comparison; otherwise the claimed SOTA margins in Tables XIII-XV are not established.
  2. [Section III.A] The knowledge distillation procedure is never described. The paper says knowledge is transferred from "the full-size teacher model" to the lightweight student, but it does not specify the teacher model (presumably PRF=1), the distillation loss function (e.g., soft-target KL divergence, feature matching), the temperature, the loss weights, or the training schedule. The "Distil" entries in Tables I-III and X-XII are therefore irreproducible, and one cannot determine whether the improvements are due to distillation or to re-initialization or retraining. A complete specification of the distillation method is required.
  3. [Section III.A (Tables I-III) and Section III.B (Tables IV-VI)] The best PRF and the best edge detector are chosen per dataset based on the test-set results (e.g., PRF=3 and Canny for SYDNEY; PRF=2 and Laplacian for UCM and RSICD). This constitutes test-set hyperparameter selection, which biases the comparison in favor of the proposed method because the baselines use fixed configurations. The authors should either select these hyperparameters on a held-out validation split and report the chosen values, or adopt a nested cross-validation scheme, to ensure an unbiased comparison.
  4. [Throughout experimental section] No error bars, standard deviations, or significance tests are reported. On the SYDNEY dataset with only 58 test images, the differences between KD-EBF and the strongest baselines (e.g., BLEU-4 0.6498 vs 0.5786 for TextGCN in Table XIII) could easily fall within sampling noise. The paper should report mean plus or minus the standard deviation over at least three random seeds for all experiments and include a paired significance test (e.g., bootstrap) for the head-to-head comparisons in Tables XIII-XV.
minor comments (6)
  1. [Section IV.C] The heading "Peformance Metrices Used" contains typos; it should be "Performance Metrics Used".
  2. [Section IV.C.4] The abbreviation "CIDE" should be "CIDEr".
  3. [Section V.E and Figure 4] The method is called "KD-EBD" in these places, but "KD-EBF" is used throughout the rest of the paper; the name should be consistent.
  4. [Tables XIII-XV] The first column header "Edge" appears to be a copy-paste error and should be "Method" (or "Model").
  5. [Section III.A] The list of PRF modifications is clear, but the paper does not specify which components are excluded from reduction (e.g., the initial patch embedding) or how the interpolation-based weight resizing is implemented in detail; please provide these for reproducibility.
  6. [Section III.B] The exact parameters of the edge detectors (e.g., Canny thresholds, Sobel or Laplacian kernel sizes) are not given; please specify them for reproducibility.

Circularity Check

0 steps flagged · score 2.0 of 10

No definitional circularity found; the paper is an empirical engineering study whose claims rest on held-out test scores and component ablations, with a benchmark-provenance caveat that is a correctness risk rather than a circular reduction.

full rationale

The paper contains no closed-form derivation whose output is equivalent to its input by construction. The three technical contributions are empirical: (1) the lightweight encoder is obtained by a PRF-based parameter reduction, and its effect is shown in Tables I to III; (2) knowledge distillation is evaluated by comparing bare versus distilled models over the same test splits; and (3) the edge-fusion module is ablated in Tables X to XII by adding edge channels to a fixed training setup. No fitted parameter is renamed as a prediction: the PRF is a model-size hyperparameter and the reported numbers are held-out test metrics, not quantities forced by the definition of the model. The only self-referential elements are the authors' own prior work: the corrected datasets from [7] and the TextGCN [7] and Conv-GPT [6] baselines in Tables XIII to XV. Using one's own previous models as baselines is a normal empirical practice and is not circular, because beating them is not guaranteed by construction. The real weakness is a comparison-protocol mismatch: Section IV.B states 'However, we have used the modified version of these datasets [7]', while the baseline scores in Tables XIII to XV come from papers that evaluated on the original, uncorrected captions. Because BLEU, METEOR, ROUGE-L, and CIDEr all measure overlap with reference captions, changing the reference set changes every metric even for identical outputs. That is a correctness and reproducibility risk for the headline SOTA claim, but it is not a self-definitional or circular step in the paper's derivation chain. No load-bearing 'uniqueness theorem' is imported from the authors' prior work, and no ansatz is smuggled in via citation. The score of 2 reflects the presence of minor, non-load-bearing self-citations and dataset provenance from the authors' own prior work, without constituting circularity.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central claim rests on three implicit assumptions: that corrected captions are a fair benchmark for comparing against published baselines, that interpolation-based weight resizing provides a useful initialization, and that an unspecified distillation procedure improves the student. No new physical or conceptual entities are introduced.

free parameters (2)
  • PRF (Parameter Reduction Factor) = 3 for SYDNEY, 2 for UCM and RSICD
    Chosen per dataset by comparing test-set metrics in Tables I-III; the reported KD-EBF uses the best-performing PRF for each dataset.
  • Edge detector choice = Canny for SYDNEY, Laplacian for UCM and RSICD
    Selected based on test-set comparisons in Tables IV-VI; differences between detectors are small.
assumptions (4)
  • domain assumption The corrected datasets from [7] are a valid and comparable benchmark to the original datasets used by cited baselines.
    Section IV.B states the authors use the modified datasets, while Tables XIII-XV report baseline scores from earlier papers that used the original captions; no re-evaluation of baselines on the corrected data is described.
  • domain assumption Interpolation-based resizing of pretrained weights preserves useful features for training the reduced encoder.
    Section III.A proposes multi-dimensional interpolation to resize weights but offers no analysis showing this initialization retains the information needed for convergence or performance.
  • ad hoc to paper Knowledge distillation improves the student model despite the procedure being unspecified.
    Tables I-III label some models 'Distil' and the abstract claims distillation helps, but Section III.A gives no loss function, temperature, teacher outputs, or training scheme, so the improvement is asserted rather than demonstrated.
  • domain assumption Edge maps concatenated as extra input channels provide complementary information that improves caption quality.
    The motivation in Section I and III.B is plausible, but the only support is the paper's own ablation tables, and no analysis isolates when edges help.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Novel Lightweight Transformer with Edge-Aware Fusion for Remote Sensing Image Captioning." pith.science (2026). https://pith.science/paper/LIS4TZCH

@misc{pith2026250609429,
  author       = {Pith},
  title        = {Pith review of: A Novel Lightweight Transformer with Edge-Aware Fusion for Remote Sensing Image Captioning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LIS4TZCH}},
  note         = {Machine review of arXiv:2506.09429}
}
read the original abstract

Transformer-based models have achieved strong performance in remote sensing image captioning by capturing long-range dependencies and contextual information. However, their practical deployment is hindered by high computational costs, especially in multi-modal frameworks that employ separate transformer-based encoders and decoders. In addition, existing remote sensing image captioning models primarily focus on high-level semantic extraction while often overlooking fine-grained structural features such as edges, contours, and object boundaries. To address these challenges, a lightweight transformer architecture is proposed by reducing the dimensionality of the encoder layers and employing a distilled version of GPT-2 as the decoder. A knowledge distillation strategy is used to transfer knowledge from a more complex teacher model to improve the performance of the lightweight network. Furthermore, an edge-aware enhancement strategy is incorporated to enhance image representation and object boundary understanding, enabling the model to capture fine-grained spatial details in remote sensing images. Experimental results demonstrate that the proposed approach significantly improves caption quality compared to state-of-the-art methods.

Figures

Figures reproduced from arXiv: 2506.09429 by the authors.

Figure 1
Figure 1. Architecture of the Proposed Model high-quality human-annotated datasets to advance the training and evaluation of large vision-language models in remote sensing image captioning. Zhan et al. [2025] [19] developed SkyEyeGPT, a multimodal large language model trained on the large-scale SkyEye-968k RS instruction-following dataset, achieving strong performance on various RS vision-language tasks without extra encoding… view at source ↗
Figure 2
Figure 2. PRF vs Total Parameters between the Parameter Reduction Factor (PRF) and the total number of parameters in the ConvNext-based transformer encoder is shown in [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Results of Different Edge Detection Techniques [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Examples of RS Image Captioning by Different Methods [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 17 canonical work pages

  1. [7]

    A textgcn-based decoding approach for improv- ing remote sensing image captioning,

    S. Das and R. Sharma, “A textgcn-based decoding approach for improv- ing remote sensing image captioning,”IEEE Geoscience and Remote Sensing Letters, 2024

    2024

  2. [1]

    Deep semantic understanding of high resolution remote sensing image,

    B. Qu, X. Li, D. Tao, and X. Lu, “Deep semantic understanding of high resolution remote sensing image,” in2016 International conference on computer, information and telecommunication systems (Cits). IEEE, 2016, pp. 1–5

    work page 2016

  3. [2]

    Exploring models and data for remote sensing image caption generation,

    X. Lu, B. Wang, X. Zheng, and X. Li, “Exploring models and data for remote sensing image caption generation,”IEEE Transactions on Geoscience and Remote Sensing, vol. 56, no. 4, pp. 2183–2195, 2017

    2017

  4. [3]

    A new cnn-rnn framework for remote sensing image captioning,

    G. Hoxha, F. Melgani, and J. Slaghenauffi, “A new cnn-rnn framework for remote sensing image captioning,” in2020 Mediterranean and Middle-East Geoscience and Remote Sensing Symposium (M2GARSS). IEEE, 2020, pp. 1–4

    work page 2020

  5. [4]

    A novel svm-based decoder for remote sens- ing image captioning,

    G. Hoxha and F. Melgani, “A novel svm-based decoder for remote sens- ing image captioning,”IEEE Transactions on Geoscience and Remote Sensing, vol. 60, pp. 1–14, 2021

    2021

  6. [5]

    Trtr-cmr: Cross- modal reasoning dual transformer for remote sensing image captioning,

    Y . Wu, L. Li, L. Jiao, F. Liu, X. Liu, and S. Yang, “Trtr-cmr: Cross- modal reasoning dual transformer for remote sensing image captioning,” IEEE Transactions on Geoscience and Remote Sensing, 2024

    2024

  7. [6]

    Good representation, better explanation: Role of convolutional neural networks in transformer-based remote sensing image captioning,

    S. Das, S. Gupta, K. Kumar, and R. Sharma, “Good representation, better explanation: Role of convolutional neural networks in transformer-based remote sensing image captioning,”arXiv preprint arXiv:2502.16095, 2025

    arXiv 2025

  8. [8]

    Learning consensus- aware semantic knowledge for remote sensing image captioning,

    Y . Li, X. Zhang, X. Cheng, X. Tang, and L. Jiao, “Learning consensus- aware semantic knowledge for remote sensing image captioning,”Pat- tern Recognition, vol. 145, p. 109893, 2024

    work page 2024

Show all 33 references
  1. [9]

    Cd4c: Change detection for remote sensing image change captioning,

    X. Li, B. Sun, Z. Wu, S. Li, and H. Guo, “Cd4c: Change detection for remote sensing image change captioning,”IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2025

    2025

  2. [10]

    Semantic-spatial feature fusion with dynamic graph refinement for remote sensing image captioning,

    M. Liu, J. Liu, and X. Zhang, “Semantic-spatial feature fusion with dynamic graph refinement for remote sensing image captioning,”arXiv preprint arXiv:2503.23453, 2025

    2025 arXiv

  3. [11]

    A residual network and bi-directional lstm based hybrid approach to remote sensing image captioning,

    M. R. Sree, M. Siddhartha, P. V . V . Reddy, R. P. Singhet al., “A residual network and bi-directional lstm based hybrid approach to remote sensing image captioning,”Procedia Computer Science, vol. 258, pp. 88–97, 2025

    2025

  4. [12]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,”Advances in neural information processing systems, vol. 30, 2017

    2017

  5. [13]

    Cross-modal retrieval and semantic refinement for remote sensing image captioning,

    Z. Li, W. Zhao, X. Du, G. Zhou, and S. Zhang, “Cross-modal retrieval and semantic refinement for remote sensing image captioning,”Remote Sensing, vol. 16, no. 1, p. 196, 2024

    2024

  6. [14]

    A multiscale group- ing transformer with clip latents for remote sensing image captioning,

    L. Meng, J. Wang, R. Meng, Y . Yang, and L. Xiao, “A multiscale group- ing transformer with clip latents for remote sensing image captioning,” IEEE Transactions on Geoscience and Remote Sensing, 2024

    2024

  7. [15]

    Bootstrapping interactive image–text alignment for remote sensing image captioning,

    C. Yang, Z. Li, and L. Zhang, “Bootstrapping interactive image–text alignment for remote sensing image captioning,”IEEE Transactions on Geoscience and Remote Sensing, vol. 62, pp. 1–12, 2024

    2024

  8. [16]

    Exploring region features in remote sensing image captioning,

    K. Zhao and W. Xiong, “Exploring region features in remote sensing image captioning,”International Journal of Applied Earth Observation and Geoinformation, vol. 127, p. 103672, 2024

    2024

  9. [17]

    Ragcap: Retrieval-augmented generation for style-aware remote sensing image captioning without fine- tuning,

    Y . Bazi, M. M. Al Rahhal, and M. Zuair, “Ragcap: Retrieval-augmented generation for style-aware remote sensing image captioning without fine- tuning,”Authorea Preprints, 2025

    2025

  10. [18]

    Rsgpt: A remote sensing vision language model and benchmark,

    Y . Hu, J. Yuan, C. Wen, X. Lu, Y . Liu, and X. Li, “Rsgpt: A remote sensing vision language model and benchmark,”ISPRS Journal of Photogrammetry and Remote Sensing, vol. 224, pp. 272–286, 2025

    2025

  11. [19]

    Skyeyegpt: Unifying remote sensing vision-language tasks via instruction tuning with large language model,

    Y . Zhan, Z. Xiong, and Y . Yuan, “Skyeyegpt: Unifying remote sensing vision-language tasks via instruction tuning with large language model,” ISPRS Journal of Photogrammetry and Remote Sensing, vol. 221, pp. 64–77, 2025

    2025

  12. [20]

    A computational approach to edge detection,

    J. Canny, “A computational approach to edge detection,”IEEE Transac- tions on pattern analysis and machine intelligence, no. 6, pp. 679–698, 1986

    1986

  13. [21]

    A 3x3 isotropic gradient operator for image processing,

    I. Sobel, G. Feldmanet al., “A 3x3 isotropic gradient operator for image processing,”a talk at the Stanford Artificial Project in, vol. 1968, pp. 271–272, 1968

    1968

  14. [22]

    Theory of edge detection,

    D. Marr and E. Hildreth, “Theory of edge detection,”Proceedings of the Royal Society of London. Series B. Biological Sciences, vol. 207, no. 1167, pp. 187–217, 1980

    1980

  15. [23]

    Saliency-guided unsupervised feature learning for scene classification,

    F. Zhang, B. Du, and L. Zhang, “Saliency-guided unsupervised feature learning for scene classification,”IEEE Transactions on Geoscience and Remote Sensing, vol. 53, no. 4, pp. 2175–2184, 2014

    2014

  16. [24]

    Bag-of-visual-words and spatial extensions for land-use classification,

    Y . Yang and S. Newsam, “Bag-of-visual-words and spatial extensions for land-use classification,” inProceedings of the 18th SIGSPATIAL in- ternational conference on advances in geographic information systems, 2010, pp. 270–279

    2010

  17. [25]

    Aid: A benchmark data set for performance evaluation of aerial scene classification,

    G.-S. Xia, J. Hu, F. Hu, B. Shi, X. Bai, Y . Zhong, L. Zhang, and X. Lu, “Aid: A benchmark data set for performance evaluation of aerial scene classification,”IEEE Transactions on Geoscience and Remote Sensing, vol. 55, no. 7, pp. 3965–3981, 2017

    2017

  18. [26]

    Relevance of unsupervised metrics in task-oriented dialogue for evaluating natural language generation,

    S. Sharma, L. El Asri, H. Schulz, and J. Zumer, “Relevance of unsupervised metrics in task-oriented dialogue for evaluating natural language generation,”CoRR, vol. abs/1706.09799, 2017. [Online]. Available: http://arxiv.org/abs/1706.09799

    2017 arXiv

  19. [27]

    Bleu: a method for automatic evaluation of machine translation,

    K. Papineni, S. Roukos, T. Ward, and W.-J. Zhu, “Bleu: a method for automatic evaluation of machine translation,” inProceedings of the 40th Annual Meeting of the Association for Computational Linguistics. Philadelphia, Pennsylvania, USA: Association for Computational Linguisti...

    2002

  20. [28]

    METEOR: An automatic metric for MT evaluation with high levels of correlation with human judgments,

    A. Lavie and A. Agarwal, “METEOR: An automatic metric for MT evaluation with high levels of correlation with human judgments,” inProceedings of the Second Workshop on Statistical Machine Translation. Prague, Czech Republic: Association for Computational Linguistics, Jun. 2007,...

    2007

  21. [29]

    ROUGE: A package for automatic evaluation of summaries,

    C.-Y . Lin, “ROUGE: A package for automatic evaluation of summaries,” inText Summarization Branches Out. Barcelona, Spain: Association for Computational Linguistics, Jul. 2004, pp. 74–81. [Online]. Available: https://aclanthology.org/W04-1013

    2004

  22. [30]

    Cider: Consensus- based image description evaluation,

    R. Vedantam, C. Lawrence Zitnick, and D. Parikh, “Cider: Consensus- based image description evaluation,” inProceedings of the IEEE confer- ence on computer vision and pattern recognition, 2015, pp. 4566–4575

    2015

  23. [31]

    Semantic descriptions of high-resolution remote sensing images,

    B. Wang, X. Lu, X. Zheng, and X. Li, “Semantic descriptions of high-resolution remote sensing images,”IEEE Geoscience and Remote Sensing Letters, vol. 16, no. 8, pp. 1274–1278, 2019

    2019

  24. [32]

    Sd-rsic: Summarization-driven deep remote sensing image captioning,

    G. Sumbul, S. Nayak, and B. Demir, “Sd-rsic: Summarization-driven deep remote sensing image captioning,”IEEE Transactions on Geo- science and Remote Sensing, vol. 59, no. 8, pp. 6922–6934, 2020

    2020

  25. [33]

    Improving image captioning systems with postprocessing strategies,

    G. Hoxha, G. Scuccato, and F. Melgani, “Improving image captioning systems with postprocessing strategies,”IEEE Transactions on Geo- science and Remote Sensing, vol. 61, pp. 1–13, 2023

    2023

Pith tools

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