Pith. sign in

REVIEW 4 major objections 4 minor 37 references

Knowledge Transfer and Domain Adaptation for Fine-Grained Remote Sensing Image Segmentation

T0 review · 4 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A CNN-based segmenter trained to mimic a frozen vision transformer's features, then refined by domain adaptation, outperforms all baselines on fine-grained grass and cloud segmentation, with reported gains of 2.57 and 3.73 mIoU.

desk verdict A plausible distillation-plus-modulation recipe with a genuinely new grass dataset, but the headline 2.57 mIoU gain is not yet trustworthy because the same test number appears to have been used to select the FMM depth. read the letter →

arxiv 2412.06664 v3 pith:7XBNM5PE submitted 2024-12-09 cs.CV

Shun Zhang , Xuechao Zou , Kai Li , Congyan Lang , Shiying Wang , Pin Tao , Tengfei Cao This is my paper · ORCID
classification cs.CV
keywords fine-grainedsegmentationremotesensingimageknowledgetransferdomainadaptationvisiontransformerfeaturealignmentgrasscoveragecloud
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 establish that a CNN-based segmentation model can be made more accurate on fine-grained remote sensing tasks by distilling feature knowledge from a frozen vision transformer encoder and then modulating those features to the target domain. It introduces the Feature Alignment Module (FAM) and Feature Modulation Module (FMM); FAM aligns multi-scale backbone features to the encoder's channel dimensions and spatial resolution, and FMM passes the aligned features through several transformer blocks before decoding. On a newly introduced five-level grass dataset and an existing cloud dataset, the method reports mIoU of 50.86 and 51.49, outperforming all compared baselines by 2.57 and 3.73 mIoU respectively. If the claim holds, it offers a practical route to strong fine-grained segmentation when labeled remote sensing data is scarce, by reusing general-purpose pretrained features.

What carries the argument

The Feature Alignment Module (FAM) and Feature Modulation Module (FMM). FAM applies 1x1 convolutions to adjust each backbone feature's channel count, bilinear interpolation to match the vision transformer's spatial resolution, and then supervised alignment via MSE and KL divergence losses. FMM applies N transformer blocks to the aligned features to shift their distribution toward the target domain, followed by a primary decoder head and an auxiliary head; the default configuration uses N=4 and loss weights λmse=λkl=0.5, λce=1.0, λaux=0.4. The combination is what carries the argument: FAM supplies transferable general features, FMM supplies domain-specific refinement, and the ablation study attributes most of the gain to the KL term and to the multi-scale alignment.

What would settle it

Take the exact reported settings and rerun the method with a separate validation split to select N and the loss weights, then evaluate on the test split once; if the mIoU margins over baselines shrink below 2.57 on grass or 3.73 on cloud, the claim that FAM and FMM cause the improvement is falsified. A second check: replace the frozen vision transformer teacher with a randomly initialized encoder and keep everything else; if the gain does not disappear, knowledge transfer is not the operative mechanism.

Watch

Extended reading notes

Core claim

The central claim is that combining knowledge transfer with domain adaptation produces state-of-the-art fine-grained segmentation in low-data remote sensing settings. The knowledge transfer step teaches a CNN backbone to mimic the multiscale features of a frozen vision transformer encoder through a weighted loss of MSE and KL divergence, after channel and spatial alignment; the domain adaptation step then refines the transferred features with a stack of transformer blocks and decodes them with a dual-head design. Evidence for the claim is the reported margin over baselines: on the grass dataset the method reaches 50.86 mIoU against 48.29 for the best baseline, and on the cloud dataset 51.49 mIoU against 43.51. The paper also contributes the fine-grained grass dataset itself, with five coverage levels labeled from satellite imagery.

Load-bearing premise

The load-bearing premise is that the test split was never used to select the model configuration; specifically, the ablation choosing N=4 transformer blocks reports the same mIoU as the final result, so if the test set influenced that choice, the central performance claim collapses.

Editorial extensions

If this is right

  • A CNN-based segmenter can outperform transformer-based segmenters on small fine-grained remote sensing datasets when it distills knowledge from a frozen transformer encoder.
  • The KL-divergence term is the single most important loss component: removing it lowers grass mIoU by 7.13 points, so the mechanism depends on distribution-level matching, not just feature regression.
  • Multi-scale feature alignment contributes more than a single-scale version, and the auxiliary head adds a smaller but consistent gain, implying the architecture's benefit is not merely extra parameters.
  • On the cloud dataset the reported mIoU of 51.49 exceeds the best baseline by 7.98 points, suggesting the recipe transfers across a second fine-grained domain.
  • The new five-level grass dataset, with coverage thresholds from under 10% to over 75%, provides a benchmark for fine-grained grassland mapping where previous datasets offer only coarse labels.

Reading between the lines

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

  • One implication the paper leaves implicit is that the same frozen-teacher-plus-modulator recipe could plausibly extend to other fine-grained remote sensing categories such as crop type, burned area, or wetland classes whenever a general-purpose vision transformer is available; the paper does not test this.
  • The very small gain from the auxiliary head relative to the KL term hints that the bottleneck in low-data fine-grained segmentation is feature adaptation rather than decoder capacity; a cheap way to test this would be to fix the decoder and vary only the number of transformer blocks.
  • Because the grass dataset is built from only four satellite images split into 1,151 patches, the reported margins may not yet reflect diversity across regions and seasons; a natural next step would be to add multi-site imagery and check whether the gains persist.
  • One could also ablate the frozen teacher by replacing it with a randomly initialized encoder; if the gain does not disappear, the mechanism is genuine knowledge transfer rather than the added modulation parameters.
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

4 major / 4 minor

Summary. The paper proposes an end-to-end framework for fine-grained remote sensing image segmentation that combines knowledge transfer from a frozen vision transformer (DINOv2) to a CNN-based student backbone via a Feature Alignment Module (FAM) and a Feature Modulation Module (FMM) for domain adaptation. The authors introduce a new fine-grained grass segmentation dataset and evaluate their method on that dataset and on a cloud segmentation dataset, reporting improvements of 2.57 mIoU on grass and 3.73 mIoU on cloud relative to various baselines. The central claim is that combining feature-alignment distillation with transformer-based feature modulation yields state-of-the-art fine-grained segmentation performance.

Significance. The proposed idea of using a frozen, general-domain vision transformer as a teacher for a lightweight CNN student, together with a small transformer-based domain-adaptation module, is timely and practically relevant for remote sensing tasks with limited labeled data. The new fine-grained grass dataset, if released with proper documentation, is a potentially valuable community resource. The method itself is simple and the ablation study attempts to isolate the contributions of the loss terms and modules. However, the current experimental evidence is weakened by evaluation-protocol issues and reporting inconsistencies, so the strength of the claimed improvements cannot be assessed from the manuscript as written.

major comments (4)
  1. [IV-C3, Tables IV and V] The number of transformer blocks N in the FMM is selected by sweeping N and reporting mIoU on what appears to be the same test split that is later used for the final comparison in Table V (both report 50.86 for N=4). Because Section IV-A1 describes only an 8:2 train/test split for the grass dataset and does not mention a validation split, the headline improvement of 2.57 mIoU over SegFormer is not protected against selection on the test set. The authors should either introduce a validation split for model selection or report results from repeated runs with error bars to rule out test-set overfitting as the source of the gain.
  2. [IV-D, Table VI, Abstract] The text states that 'the previous best method, HRCloudNet' achieves a mIoU of 43.51, but Table VI lists UNetMobV2 at 47.76, which is higher, and the Abstract's claimed 3.73 mIoU improvement is computed relative to UNetMobV2 (51.49 - 47.76), not relative to the stated best baseline. This internal inconsistency makes the reported comparison against the best cloud baseline unreliable and should be corrected and clarified.
  3. [III-A, IV-B] The CNN-based backbone used throughout the experiments is never named. The method is defined as aligning features from a 'CNN-based backbone' with a frozen vision transformer encoder, but without specifying the backbone architecture (e.g., ResNet-50 or ResNet-101), the FAM/FMM contribution cannot be isolated, the comparisons against baselines are not reproducible, and the claim that gains come from knowledge transfer rather than from a particular backbone cannot be assessed. Please provide the exact backbone configuration.
  4. [All experimental tables, Abstract] All reported metrics are single-run point estimates without error bars, repeated runs, or significance tests, so the word 'significant' in the Abstract is not supported statistically. This is especially important because the grass improvement over SegFormer is only 2.57 mIoU, and the ablation curve in Table IV is non-monotonic in N; the authors should report variance or at least multiple seeds to substantiate the claimed improvements.
minor comments (4)
  1. [IV-C1 vs IV-C3, Tables II, IV, V] The best mIoU reported in the loss ablation (50.81, Table II) differs from the final configuration's mIoU (50.86, Tables IV and V) without an explanation of whether the FMM depth or other settings differ between these tables; please clarify the exact configurations.
  2. [III-A2] The term 'kullback-leibler' should be capitalized as 'Kullback-Leibler'.
  3. [IV-B2] The statement that 'All of feature extraction networks are pre-trained on the ImageNet' is inaccurate for DINOv2, which is self-supervised on LVD-142M rather than trained on ImageNet labels; please correct this and specify the exact DINOv2 variant used (e.g., ViT-S/14 or ViT-B/14).
  4. [IV-A1] The new grass dataset is a key contribution, but the paper does not state where the dataset will be released or under what license; please add availability details.

Circularity Check

1 steps flagged · score 6.0 of 10

Grass-dataset headline gain is the FMM-depth selection value reported as an independent test result; cloud claim and method derivation are otherwise self-contained.

  1. fitted input called prediction [Section IV-A1 (dataset split), Section IV-C3 (FMM ablation), Table V]
    "For the grass dataset: “final dataset comprises 1,151 pairs of 256×256 patches, split into training and testing sets with an 8:2 ratio.” Section IV-C3: “As the number of Transformer Blocks in FMM increases, with the best results achieved at N=4, reaching mIoU of 50.86%.” Table V: “Ours 50.86 74.26 65.01.”"

    The grass dataset has no validation split, so the table used to select the FMM depth N is computed on the test split. The value that maximizes the selection criterion (N=4, mIoU 50.86 in Table IV) is then reported verbatim as the method’s final test performance (Table V: Ours 50.86). Consequently the paper’s headline improvement over SegFormer (50.86 − 48.29 = 2.57 mIoU) is not an unbiased prediction of the method’s performance; it is the maximum of the five configurations tried on the evaluation set. The reported gain is therefore forced, in part, by the selection rule, i.e. the “prediction” is the fitted quantity renamed as a result. The cloud result (external dataset, 6:2:2 split) and the loss-component ablations are separate and not affected by this specific circularity.

full rationale

The method itself is not definitionally circular: FAM/FMM are defined independently of the reported metrics, the losses (Lkt, Lda) are standard and do not encode the target mIoU, and the comparison baselines are external. The only self-citation ([25], used as a general pointer for domain adaptation) is not load-bearing. However, the grass-dataset evaluation protocol makes the central numeric claim circular in a narrower sense: the same 8:2 split is used for FMM-depth selection and final reporting, and the chosen N=4 score from Table IV is identical to the final “Ours” score in Table V. Thus the 2.57 mIoU headline margin is not an independent test-set estimate but the result of selecting the best of several configurations on the evaluation split. The cloud result (51.49 vs. 47.76 UNetMobv2, 3.73 mIoU) uses a proper 6:2:2 split and external dataset, and the qualitative comparisons are independent. Overall: partial circularity confined to the grass claim, not a derivation-level equivalence.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on a handful of domain assumptions: DINOv2 features help, trainable transformer blocks can adapt domains, labels are reliable, and the test set was not used for tuning. The hand-chosen loss weights and the FMM depth N are free parameters that the reported gains depend on. No fundamentally new physical entities are introduced.

free parameters (3)
  • Loss weights (lambda_mse, lambda_kl, lambda_aux, lambda_ce, lambda_kt, lambda_da) = 0.5, 0.5, 0.4, 1.0, 1.0, 1.0
    Hand-chosen defaults in Eqs. (5), (8), and (9); no sensitivity analysis is provided, so the reported gains are conditional on these choices.
  • Number of FMM transformer blocks N = 4
    Chosen from the ablation in Table IV; performance is non-monotonic (N=3: 49.64, N=2: 50.57) and no error bars are given, so the choice may reflect noise or test-set tuning.
  • Number of feature scales n = 4
    Set as the default in Section III-A1; the ablation in Table III shows multi-scale helps, but the exact scales are not specified.
assumptions (4)
  • domain assumption Frozen DINOv2 encoder features are a beneficial knowledge source for fine-grained remote sensing segmentation.
    Invoked in Section III-A; the entire knowledge transfer loss assumes these features can be distilled into the CNN to improve segmentation.
  • domain assumption Transformer blocks in FMM can adapt general-domain features to the remote sensing target domain.
    Invoked in Section III-B; no unlabeled target data or adversarial objective is used, so 'domain adaptation' is just additional trainable layers.
  • domain assumption The grass dataset labels are accurate enough to train and evaluate on.
    Section IV-A1 describes labeling with X-AnyLabeling and manual refinement, but no inter-annotator agreement or label quality metrics are reported.
  • domain assumption The train/test split is independent and was not used for model selection.
    Section IV-A1 gives an 8:2 split with no validation set; this assumption is load-bearing for the validity of the reported test metrics.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Knowledge Transfer and Domain Adaptation for Fine-Grained Remote Sensing Image Segmentation." pith.science (2026). https://pith.science/paper/7XBNM5PE

@misc{pith2026241206664,
  author       = {Pith},
  title        = {Pith review of: Knowledge Transfer and Domain Adaptation for Fine-Grained Remote Sensing Image Segmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7XBNM5PE}},
  note         = {Machine review of arXiv:2412.06664}
}
read the original abstract

Fine-grained remote sensing image segmentation is essential for accurately identifying detailed objects in remote sensing images. Recently, vision transformer models (VTMs) pre-trained on large-scale datasets have demonstrated strong zero-shot generalization. However, directly applying them to specific tasks may lead to domain shift. We introduce a novel end-to-end learning paradigm combining knowledge guidance with domain refinement to enhance performance. We present two key components: the Feature Alignment Module (FAM) and the Feature Modulation Module (FMM). FAM aligns features from a CNN-based backbone with those from the pretrained VTM's encoder using channel transformation and spatial interpolation, and transfers knowledge via KL divergence and L2 normalization constraint. FMM further adapts the knowledge to the specific domain to address domain shift. We also introduce a fine-grained grass segmentation dataset and demonstrate, through experiments on two datasets, that our method achieves a significant improvement of 2.57 mIoU on the grass dataset and 3.73 mIoU on the cloud dataset. The results highlight the potential of combining knowledge transfer and domain adaptation to overcome domain-related challenges and data limitations. The project page is available at https://xavierjiezou.github.io/KTDA/.

Figures

Figures reproduced from arXiv: 2412.06664 by the authors.

Figure 1
Figure 1. Overview of the proposed framework that integrates knowledge transfer and domain adaptation for fine-grained remote sensing image segmentation. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Detailed structure of the FAM and FMM. where Lkt denotes the knowledge transfer loss, and we use λmse = λkl = 0.5 as default weights. B. Domain Adaptation Domain adaptation (see Fig. 1B) involves a feature modu￾lation module built upon transformer blocks to modulate the feature distribution from the source (VTM) to target (seg￾mentation) domains. The module processes the input features F ′′ i through N sequential tr… view at source ↗
Figure 3
Figure 3. Comparison of visualization segmentation results of different models on the fine-grained grass and cloud segmentation datasets. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 36 canonical work pages

  1. [1]

    Large-scale land cover mapping with fine-grained classes via class-aware semi- supervised semantic segmentation,

    Runmin Dong, Lichao Mou, and Mengxuan et al. Chen, “Large-scale land cover mapping with fine-grained classes via class-aware semi- supervised semantic segmentation,” in ICCV, 2023, pp. 16783–16793

    work page 2023

  2. [2]

    Fine-grained recognition for oriented ship against complex scenes in optical remote sensing images,

    Yaqi Han, Xinyi Yang, Tian Pu, and Zhenming Peng, “Fine-grained recognition for oriented ship against complex scenes in optical remote sensing images,” TGRS, vol. 60, pp. 1–18, 2021

    work page 2021

  3. [3]

    Multiattention network for semantic segmentation of fine-resolution remote sensing images,

    Rui Li, Shunyi Zheng, Ce Zhang, Chenxi Duan, Jianlin Su, Libo Wang, and Peter M Atkinson, “Multiattention network for semantic segmentation of fine-resolution remote sensing images,” TGRS, vol. 60, pp. 1–13, 2021

    work page 2021

  4. [4]

    Fully convolu- tional networks for semantic segmentation,

    Jonathan Long, Evan Shelhamer, and Trevor Darrell, “Fully convolu- tional networks for semantic segmentation,” in CVPR, 2015, pp. 3431– 3440

    work page 2015

  5. [5]

    U-net: Con- volutional networks for biomedical image segmentation,

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox, “U-net: Con- volutional networks for biomedical image segmentation,” in MICCAI. Springer, 2015, pp. 234–241

    work page 2015

  6. [6]

    Pyramid scene parsing network,

    Hengshuang Zhao, Jianping Shi, Xiaojuan Qi, Xiaogang Wang, and Jiaya Jia, “Pyramid scene parsing network,” in CVPR, 2017, pp. 2881–2890

    work page 2017

  7. [7]

    Encoder-decoder with atrous separable convolution for semantic image segmentation,

    Liang-Chieh Chen, Yukun Zhu, George Papandreou, Florian Schroff, and Hartwig Adam, “Encoder-decoder with atrous separable convolution for semantic image segmentation,” in ECCV, 2018, pp. 801–818

    work page 2018

  8. [8]

    PMAA: A progressive multi-scale attention autoencoder model for high-performance cloud removal from multi-temporal satellite imagery,

    Xuechao Zou, Kai Li, Junliang Xing, Pin Tao, and Yachao Cui, “PMAA: A progressive multi-scale attention autoencoder model for high-performance cloud removal from multi-temporal satellite imagery,” in ECAI, 2023, vol. 372, pp. 3165–3172

    work page 2023

Show all 37 references
  1. [9]

    Diffcr: A fast conditional diffusion framework for cloud removal from optical satellite images,

    Xuechao Zou, Kai Li, Junliang Xing, Yu Zhang, Shiying Wang, Lei Jin, and Pin Tao, “Diffcr: A fast conditional diffusion framework for cloud removal from optical satellite images,” TGRS, vol. 62, pp. 1–14, 2024

    2024

  2. [10]

    Sam-clip: Merging vision foundation models towards semantic and spatial understanding,

    Haoxiang Wang, Pavan Kumar Anasosalu Vasu, and Fartash et al. Faghri, “Sam-clip: Merging vision foundation models towards semantic and spatial understanding,” in CVPR, 2024, pp. 3635–3647

    2024

  3. [11]

    DI- NOv2: Learning Robust Visual Features without Supervision,

    Maxime Oquab, Timoth ´ee Darcet, and Th ´eo et al. Moutakanni, “DI- NOv2: Learning Robust Visual Features without Supervision,” TMLR, pp. 1–31, 2024

    2024

  4. [12]

    Learning transferable visual models from natural language supervision,

    Alec Radford, Jong Wook Kim, and Chris et al. Hallacy, “Learning transferable visual models from natural language supervision,” in ICML, 2021, vol. 139, pp. 8748–8763

    2021

  5. [13]

    Iianet: An intra- and inter-modality attention network for audio-visual speech separation,

    Kai Li, Runxuan Yang, Fuchun Sun, and Xiaolin Hu, “Iianet: An intra- and inter-modality attention network for audio-visual speech separation,” in ICML, 2024

    2024

  6. [14]

    Imagenet large scale visual recognition challenge,

    Olga Russakovsky, Jia Deng, and Hao et al. Su, “Imagenet large scale visual recognition challenge,” IJCV, vol. 115, pp. 211–252, 2015

    2015

  7. [15]

    Segformer: Simple and efficient design for semantic segmentation with transformers,

    Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, and Ping Luo, “Segformer: Simple and efficient design for semantic segmentation with transformers,” in NeurIPS, 2021, vol. 34, pp. 12077–12090

    2021

  8. [16]

    Masked-attention mask transformer for universal image segmentation,

    Bowen Cheng, Ishan Misra, Alexander G Schwing, Alexander Kirillov, and Rohit Girdhar, “Masked-attention mask transformer for universal image segmentation,” in CVPR, 2022, pp. 1290–1299

    2022

  9. [17]

    Cdnet: Cnn-based cloud detection for remote sensing imagery,

    Jingyu Yang, Jianhua Guo, Huanjing Yue, Zhiheng Liu, Haofeng Hu, and Kun Li, “Cdnet: Cnn-based cloud detection for remote sensing imagery,” TGRS, vol. 57, no. 8, pp. 6195–6211, 2019

    2019

  10. [18]

    Cdnetv2: Cnn-based cloud detection for remote sensing imagery with cloud-snow coexistence,

    Jianhua Guo, Jingyu Yang, Huanjing Yue, Hai Tan, Chunping Hou, and Kun Li, “Cdnetv2: Cnn-based cloud detection for remote sensing imagery with cloud-snow coexistence,” TGRS, vol. 59, no. 1, pp. 700– 713, 2021

    2021

  11. [19]

    Distilling the knowledge in a neural network,

    Geoffrey Hinton, Oriol Vinyals, and Jeff Dean, “Distilling the knowledge in a neural network,” in NeurIPSW, 2014, pp. 1–9

    2014

  12. [20]

    Faster segment anything: Towards lightweight sam for mobile applications,

    Chaoning Zhang, Dongshen Han, and Yu et al. Qiao, “Faster segment anything: Towards lightweight sam for mobile applications,” arXiv preprint arXiv:2306.14289, 2023

    2023 arXiv

  13. [21]

    Efficientsam: Leveraged masked image pretraining for efficient segment anything,

    Yunyang Xiong, Bala Varadarajan, and Lemeng et al. Wu, “Efficientsam: Leveraged masked image pretraining for efficient segment anything,” in CVPR, 2024, pp. 16111–16121

    2024

  14. [22]

    Segment anything,

    Alexander Kirillov, Eric Mintun, and Nikhila et al. Ravi, “Segment anything,” in ICCV, 2023, pp. 4015–4026

    2023

  15. [23]

    An image is worth 16x16 words: Transformers for image recognition at scale,

    Alexey Dosovitskiy, Lucas Beyer, and Alexander et al. Kolesnikov, “An image is worth 16x16 words: Transformers for image recognition at scale,” in ICLR, 2020, pp. 1–21

    2020

  16. [24]

    Masked autoencoders are scalable vision learners,

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Doll ´ar, and Ross Girshick, “Masked autoencoders are scalable vision learners,” in CVPR, 2022, pp. 16000–16009

    2022

  17. [25]

    Adapting vision foundation models for robust cloud segmentation in remote sensing images,

    Xuechao Zou, Shun Zhang, Kai Li, Shiying Wang, Junliang Xing, Lei Jin, Congyan Lang, and Pin Tao, “Adapting vision foundation models for robust cloud segmentation in remote sensing images,” 2024

    2024

  18. [26]

    Eurosat: A novel dataset and deep learning benchmark for land use and land cover classification,

    Patrick Helber, Benjamin Bischke, Andreas Dengel, and Damian Borth, “Eurosat: A novel dataset and deep learning benchmark for land use and land cover classification,” IEEE J-STARS, 2019

    2019

  19. [27]

    Land-cover classification with high-resolution remote sensing images using transferable deep models,

    Xin-Yi Tong, Gui-Song Xia, Qikai Lu, Huanfeng Shen, Shengyang Li, Shucheng You, and Liangpei Zhang, “Land-cover classification with high-resolution remote sensing images using transferable deep models,” RSE, vol. 237, pp. 111322, 2020

    2020

  20. [28]

    Landcover.ai: Dataset for automatic mapping of buildings, woodlands, water and roads from aerial imagery,

    Adrian Boguszewski, Dominik Batorski, Natalia Ziemba-Jankowska, Tomasz Dziedzic, and Anna Zambrzycka, “Landcover.ai: Dataset for automatic mapping of buildings, woodlands, water and roads from aerial imagery,” in CVPR, June 2021, pp. 1102–1110

    2021

  21. [29]

    Algorithms for semantic segmentation of multispectral remote sensing imagery using deep learning,

    Ronald Kemker, Carl Salvaggio, and Christopher Kanan, “Algorithms for semantic segmentation of multispectral remote sensing imagery using deep learning,” ISPRS J. Photogramm. Remote Sens. , 2018

    2018

  22. [30]

    Cloud detec- tion algorithm comparison and validation for operational landsat data products,

    Steve Foga, Pat L Scaramuzza, and Song et al. Guo, “Cloud detec- tion algorithm comparison and validation for operational landsat data products,” RSE, vol. 194, pp. 379–390, 2017

    2017

  23. [31]

    Unified perceptual parsing for scene understanding,

    Tete Xiao, Yingcheng Liu, Bolei Zhou, Yuning Jiang, and Jian Sun, “Unified perceptual parsing for scene understanding,” in ECCV, 2018, pp. 418–434

    2018

  24. [32]

    Decoupled weight decay regulariza- tion,

    Ilya Loshchilov and Frank Hutter, “Decoupled weight decay regulariza- tion,” in ICLR, 2018

    2018

  25. [33]

    Remote sensing image cloud detection using a shallow convolutional neural network,

    Dengfeng Chai, Jingfeng Huang, Minghui Wu, Xiaoping Yang, and Ruisheng Wang, “Remote sensing image cloud detection using a shallow convolutional neural network,” ISPRS J. Photogramm. Remote Sens. , vol. 209, pp. 66–84, 2024

    2024

  26. [34]

    Mcdnet: Multilevel cloud detection network for remote sensing images based on dual-perspective change-guided and multi-scale feature fusion,

    Junwu Dong, Yanhui Wang, Yang Yang, Mengqin Yang, and Jun Chen, “Mcdnet: Multilevel cloud detection network for remote sensing images based on dual-perspective change-guided and multi-scale feature fusion,” JAG, vol. 129, pp. 103820, 2024

    2024

  27. [35]

    Cloudsen12, a global dataset for semantic understanding of cloud and cloud shadow in sentinel-2,

    Cesar Aybar, Luis Ysuhuaylas, and Jhomira et al. Loja, “Cloudsen12, a global dataset for semantic understanding of cloud and cloud shadow in sentinel-2,” Scientific data, vol. 9, no. 1, pp. 782, 2022

    2022

  28. [36]

    High-resolution cloud detection network,

    Jingsheng Li, Tianxiang Xue, Jiayi Zhao, Jingmin Ge, Yufang Min, Wei Su, and Kun Zhan, “High-resolution cloud detection network,” JEI, vol. 33, no. 4, pp. 043027–043027, 2024

    2024

  29. [37]

    Kappa- mask: Ai-based cloudmask processor for sentinel-2,

    Marharyta Domnich, Indrek S ¨unter, and Heido et al. Trofimov, “Kappa- mask: Ai-based cloudmask processor for sentinel-2,” Remote Sensing , vol. 13, no. 20, pp. 4100, 2021

    2021

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.