pith. sign in

arxiv: 2604.05527 · v1 · submitted 2026-04-07 · 💻 cs.CV

Prior-guided Fusion of Multimodal Features for Change Detection from Optical-SAR Images

Xuanguang Liu , Lei Ding , Yujie Li , Chenguang Dai , Zhenchao Zhang , Mengmeng Li , Ziyi Yang , Yifan Sun
show 2 more authors
Yongqi Sun Hanyun Wang
This is my paper

Pith reviewed 2026-05-10 19:52 UTC · model grok-4.3

classification 💻 cs.CV
keywords multimodal change detectionoptical-SAR fusionsemantic priorsfeature fusionremote sensingchange detectionSTSF-Net
0
0 comments X

The pith

STSF-Net fuses optical and SAR features using semantic priors to reduce pseudo-changes in multimodal detection.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces STSF-Net to detect changes between optical and SAR satellite images more accurately than prior methods. It extracts features unique to each sensor type to capture real semantic shifts while adding shared spatio-temporal features to filter out artifacts from differing imaging physics. A key step adaptively weights the two streams by pulling semantic guidance from pre-trained models, so the network trusts the more reliable modality at each location. The authors also release Delta-SN6, the first open very-high-resolution multiclass benchmark pairing fully polarimetric SAR with optical imagery. Tests on Delta-SN6, BRIGHT, and Wuhan-Het show consistent gains in mean intersection-over-union.

Core claim

STSF-Net jointly models modality-specific and spatio-temporal common features to enhance change representations, while an optical-SAR fusion strategy adaptively adjusts feature importance using semantic priors from pre-trained foundational models.

What carries the argument

The prior-guided adaptive fusion that weights optical and SAR features according to semantic priors extracted from pre-trained models.

If this is right

  • Modality-specific features surface genuine semantic changes while common features suppress sensor-induced false positives.
  • Adaptive weighting driven by external semantic priors improves multiclass change maps on very-high-resolution data.
  • The open Delta-SN6 dataset supplies the first public VHR fully polarimetric SAR plus optical multiclass change benchmark.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the priors generalize across sensors, the same guidance mechanism could transfer to other multimodal remote-sensing tasks such as segmentation or object detection.
  • The approach implicitly suggests that large vision models can supply stable semantic context even when the target domain is satellite imagery.
  • Future work could test whether the fusion strategy still works when the pre-trained model is trained only on optical data rather than mixed sources.

Load-bearing premise

Semantic priors taken from pre-trained foundational models stay reliable and unbiased when used to steer fusion of optical and SAR data for change detection.

What would settle it

Removing the semantic-prior guidance or swapping it for a different pre-trained model produces no mIoU gain on Delta-SN6 or BRIGHT.

Figures

Figures reproduced from arXiv: 2604.05527 by Chenguang Dai, Hanyun Wang, Lei Ding, Mengmeng Li, Xuanguang Liu, Yifan Sun, Yongqi Sun, Yujie Li, Zhenchao Zhang, Ziyi Yang.

Figure 1
Figure 1. Figure 1: The Delta-SN6 dataset provides multimodal and multi-temporal remote sensing images along with ground truth labels, [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: An overview of the proposed MMCD architecture. It contains three core components: a modal-specific feature encoder [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Details of the Modal-specific feature Extractor. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Extraction of cross-modal common features with spatio-temporal correlations. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Workflow for fusing modality-specific and common [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Results of different methods in the testing areas (a)-(d) on the Wuhan-Het dataset. [PITH_FULL_IMAGE:figures/full_fig_p010_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Results among different methods in the testing areas (a)-(d) on the BRIGHT dataset. [PITH_FULL_IMAGE:figures/full_fig_p012_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Results among different methods in the testing areas (a)-(d) on the Delta-SN6 dataset. [PITH_FULL_IMAGE:figures/full_fig_p013_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Qualitative comparison of the MMCD results obtained using different network modules on the BRIGHT and Delta [PITH_FULL_IMAGE:figures/full_fig_p014_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Comparison of feature distributions of optical and SAR before and after STCFM processing. [PITH_FULL_IMAGE:figures/full_fig_p015_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Visualized change in feature response intensity before and after PGFFM addition on BRIGHT and Delta-SN6 datasets. [PITH_FULL_IMAGE:figures/full_fig_p016_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Visualization of specific features, common features, and the fused change feature. [PITH_FULL_IMAGE:figures/full_fig_p018_12.png] view at source ↗
read the original abstract

Multimodal change detection (MMCD) identifies changed areas in multimodal remote sensing (RS) data, demonstrating significant application value in land use monitoring, disaster assessment, and urban sustainable development. However, literature MMCD approaches exhibit limitations in cross-modal interaction and exploiting modality-specific characteristics. This leads to insufficient modeling of fine-grained change information, thus hindering the precise detection of semantic changes in multimodal data. To address the above problems, we propose STSF-Net, a framework designed for MMCD between optical and SAR images. STSF-Net jointly models modality-specific and spatio-temporal common features to enhance change representations. Specifically, modality-specific features are exploited to capture genuine semantic change signals, while spatio-temporal common features are embedded to suppress pseudo-changes caused by differences in imaging mechanisms. Furthermore, we introduce an optical and SAR feature fusion strategy that adaptively adjusts feature importance based on semantic priors obtained from pre-trained foundational models, enabling semantic-guided adaptive fusion of multi-modal information. In addition, we introduce the Delta-SN6 dataset, the first openly-accessible multiclass MMCD benchmark consisting of very-high-resolution (VHR) fully polarimetric SAR and optical images. Experimental results on Delta-SN6, BRIGHT, and Wuhan-Het datasets demonstrate that our method outperforms the state-of-the-art (SOTA) by 3.21%, 1.08%, and 1.32% in mIoU, respectively. The associated code and Delta-SN6 dataset will be released at: https://github.com/liuxuanguang/STSF-Net.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper proposes STSF-Net for multimodal change detection (MMCD) between optical and SAR images. It jointly extracts modality-specific features to capture genuine semantic changes and spatio-temporal common features to suppress pseudo-changes arising from differing imaging mechanisms. A central component is an adaptive optical-SAR feature fusion module that re-weights features using semantic priors extracted from pre-trained foundational models. The authors also release the Delta-SN6 dataset (first open multiclass VHR optical-SAR MMCD benchmark) and report mIoU gains of 3.21%, 1.08%, and 1.32% over prior SOTA on Delta-SN6, BRIGHT, and Wuhan-Het, respectively, with code and data to be released.

Significance. If the reported gains prove robust and causally attributable to the prior-guided fusion, the work would offer a practical advance in handling cross-modal discrepancies in RS change detection, with direct relevance to land-use monitoring and disaster assessment. The public release of the Delta-SN6 benchmark and associated code constitutes a clear positive contribution to the community, independent of the algorithmic novelty.

major comments (2)
  1. [§3.3] §3.3 (Semantic Prior-Guided Fusion): The description of prior extraction and adaptive weighting presupposes that semantic priors transferred from general pre-trained foundational models remain accurate and unbiased on VHR SAR-optical pairs, yet no quantitative validation (e.g., prior accuracy metrics, domain-shift analysis, or error propagation study) is provided. This assumption is load-bearing for the central claim that the fusion strategy, rather than other architectural elements, drives the observed mIoU improvements.
  2. [§4] §4 (Experiments, Tables 2–4): The ablation studies do not isolate the contribution of the semantic-prior component from the modality-specific and spatio-temporal branches; consequently the small reported gains (1–3 % mIoU) cannot be confidently attributed to the proposed fusion rather than to increased model capacity or implementation details. In addition, no statistical significance tests or multi-run standard deviations are reported.
minor comments (2)
  1. [Abstract] Abstract: the phrasing 'literature MMCD approaches exhibit limitations' is grammatically awkward; rephrase to 'existing MMCD approaches in the literature exhibit limitations'.
  2. [§2] §2 (Related Work): several citations to foundational-model papers are present but lack discussion of their specific pre-training domains (natural images vs. remote-sensing), which is relevant to the domain-shift concern.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback and for recognizing the value of the Delta-SN6 benchmark. We address each major comment below with clarifications and planned revisions.

read point-by-point responses

  1. Referee: [§3.3] §3.3 (Semantic Prior-Guided Fusion): The description of prior extraction and adaptive weighting presupposes that semantic priors transferred from general pre-trained foundational models remain accurate and unbiased on VHR SAR-optical pairs, yet no quantitative validation (e.g., prior accuracy metrics, domain-shift analysis, or error propagation study) is provided. This assumption is load-bearing for the central claim that the fusion strategy, rather than other architectural elements, drives the observed mIoU improvements.

    Authors: We agree that the manuscript lacks explicit quantitative validation of the transferred semantic priors on the target VHR optical-SAR domain. The priors are obtained from general pre-trained models and used only to modulate adaptive weights within the fusion module; the network is trained end-to-end and can therefore compensate for moderate prior inaccuracies. Nevertheless, to strengthen attribution of the reported gains specifically to the prior-guided mechanism, we will add in the revision: (i) prior accuracy metrics computed on a held-out subset of optical-SAR pairs, (ii) a domain-shift analysis comparing prior quality on optical versus SAR inputs, and (iii) a sensitivity study that perturbs the priors and measures downstream mIoU change. These additions will be placed in §3.3 and an accompanying appendix. revision: yes

  2. Referee: [§4] §4 (Experiments, Tables 2–4): The ablation studies do not isolate the contribution of the semantic-prior component from the modality-specific and spatio-temporal branches; consequently the small reported gains (1–3 % mIoU) cannot be confidently attributed to the proposed fusion rather than to increased model capacity or implementation details. In addition, no statistical significance tests or multi-run standard deviations are reported.

    Authors: We concur that the current ablation tables do not isolate the semantic-prior guidance from the modality-specific and spatio-temporal branches, nor do they quantify run-to-run variability. In the revised manuscript we will augment Tables 2–4 with a dedicated ablation that disables the prior input (replacing it with uniform or learned non-prior weights while preserving parameter count) and with results averaged over five random seeds together with standard deviations. We will also report paired statistical significance tests (Wilcoxon signed-rank) between the full model and the ablated variant. These changes will allow clearer attribution of the 1–3 % mIoU gains to the prior-guided fusion component. revision: yes

Circularity Check

0 steps flagged

No circularity: new architecture with external priors and held-out empirical results

full rationale

The paper proposes STSF-Net as a novel multimodal fusion network that extracts modality-specific and spatio-temporal common features, then applies an adaptive fusion module guided by semantic priors from external pre-trained foundational models. Performance is reported via mIoU gains on three independent test datasets (Delta-SN6 introduced here, plus BRIGHT and Wuhan-Het). No equations, derivations, or 'predictions' are presented that reduce by construction to quantities fitted from the same data or defined in terms of the target outputs. No self-citation chains, uniqueness theorems, or ansatzes imported from prior author work are invoked as load-bearing justification. The central claims rest on architectural design choices and standard supervised evaluation, which are self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The approach depends on standard deep learning training assumptions plus the domain-specific assumption that pre-trained model priors transfer effectively to optical-SAR change detection.

free parameters (1)
  • Adaptive fusion parameters
    Learned or tuned weights that adjust modality importance based on priors during training.
axioms (1)
  • domain assumption Semantic priors from pre-trained foundational models are transferable to optical-SAR remote sensing images for guiding feature fusion.
    Invoked to enable the adaptive fusion strategy described in the abstract.

pith-pipeline@v0.9.0 · 5618 in / 1254 out tokens · 65483 ms · 2026-05-10T19:52:56.691107+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

63 extracted references · 63 canonical work pages

  1. [1]

    J. F. Brown, H. J. Tollerud, C. P. Barber, Q. Zhou, J. L. Dwyer, J. E. V ogelmann, T. R. Loveland, C. E. Woodcock, S. V . Stehman, Z. Zhu, B. W. Pengra, K. Smith, J. A. Horton, G. Xian, R. F. Auch, T. L. Sohl, K. L. Sayler, A. L. Gallant, D. Zelenak, R. R. Reker, and J. Rover, “Lessons learned implementing an operational continuous united states national ...

    work page 2020

  2. [2]

    Cross-modal feature interaction network for heterogeneous change detection,

    Z. Yang, X. Wang, H. Lin, M. Li, and M. Lin, “Cross-modal feature interaction network for heterogeneous change detection,”Geo-spat. Inf. Sci., vol. 28, no. 5, pp. 2358–2379, 2025

    work page 2025

  3. [3]

    Land cover change detection with hyperspectral remote sensing images: A survey,

    Z. Lv, M. Zhang, W. Sun, T. Lei, J. A. Benediktsson, and T. Liu, “Land cover change detection with hyperspectral remote sensing images: A survey,”Inf. Fusion, vol. 123, p. 103257, 2025

    work page 2025

  4. [4]

    The regularized iteratively reweighted mad method for change detection in multi- and hyperspectral data,

    A. A. Nielsen, “The regularized iteratively reweighted mad method for change detection in multi- and hyperspectral data,”IEEE Trans. Image Process., vol. 16, no. 2, pp. 463–478, 2007

    work page 2007

  5. [5]

    Unsupervised land-use change detection using multi-temporal poi embedding,

    Y . Yao, Q. Zhu, Z. Guo, W. Huang, Y . Zhang, X. Yan, A. Dong, Z. Jiang, H. Liu, and Q. Guan, “Unsupervised land-use change detection using multi-temporal poi embedding,”International Journal of Geographical Information Science, vol. 37, no. 11, pp. 2392–2415, 2023

    work page 2023

  6. [6]

    From single- to multi-modal remote sensing imagery interpretation: a survey and taxonomy,

    X. Sun, Y . Tian, W. Lu, P. Wang, R. Niu, H. Yu, and K. Fu, “From single- to multi-modal remote sensing imagery interpretation: a survey and taxonomy,”Sci. China Inf. Sci., vol. 66, no. 4, p. 140301, 2023

    work page 2023

  7. [7]

    Deep learning in multimodal remote sensing data fusion: A compre- hensive review,

    J. Li, D. Hong, L. Gao, J. Yao, K. Zheng, B. Zhang, and J. Chanussot, “Deep learning in multimodal remote sensing data fusion: A compre- hensive review,”Int. J. Appl. Earth Obs. Geoinf., vol. 112, p. 102926, 2022

    work page 2022

  8. [8]

    Multimodal sentiment analy- sis—a comprehensive survey from a fusion methods perspective,

    K. Zhao, M. Zheng, Q. Li, and J. Liu, “Multimodal sentiment analy- sis—a comprehensive survey from a fusion methods perspective,”IEEE Access, vol. 13, pp. 64 556–64 583, 2025

    work page 2025

  9. [9]

    Gccd: A generative cross-domain change detection network,

    M. Zhang, C. Guo, Y . Zhang, H. Liu, and W. Li, “Gccd: A generative cross-domain change detection network,”IEEE Trans. Geosci. Remote Sens., vol. 62, pp. 1–10, 2024

    work page 2024

  10. [10]

    Changeclip: Remote sensing change detection with multimodal vision-language representation learn- ing,

    S. Dong, L. Wang, B. Du, and X. Meng, “Changeclip: Remote sensing change detection with multimodal vision-language representation learn- ing,”ISPRS J. Photogramm. Remote Sens., vol. 208, pp. 53–69, 2024

    work page 2024

  11. [11]

    Transformer-based multimodal change detection with multitask consistency constraints,

    B. Liu, H. Chen, K. Li, and M. Y . Yang, “Transformer-based multimodal change detection with multitask consistency constraints,”Inf. Fusion, vol. 108, p. 102358, 2024

    work page 2024

  12. [12]

    Remote sens- ing spatiotemporal vision–language models: A comprehensive survey,

    C. Liu, J. Zhang, K. Chen, M. Wang, Z. Zou, and Z. Shi, “Remote sens- ing spatiotemporal vision–language models: A comprehensive survey,” IEEE Geosci. Remote Sens. Mag., pp. 2–42, 2025

    work page 2025

  13. [13]

    An approach to differentiate informal settlements using spectral, texture, geomorphology and road accessibility metrics,

    K. K. Owen and D. W. Wong, “An approach to differentiate informal settlements using spectral, texture, geomorphology and road accessibility metrics,”Appl. Geogr., vol. 38, pp. 107–118, 2013

    work page 2013

  14. [14]

    Domain separation networks,

    K. Bousmalis, G. Trigeorgis, N. Silberman, D. Krishnan, and D. Erhan, “Domain separation networks,” inProceedings of the 30th International Conference on Neural Information Processing Systems, ser. NIPS’16. Red Hook, NY , USA: Curran Associates Inc., 2016, p. 343–351

    work page 2016

  15. [15]

    Freefusion: Infrared and visible image fusion via cross reconstruction learning,

    W. Zhao, H. Cui, H. Wang, Y . He, and H. Lu, “Freefusion: Infrared and visible image fusion via cross reconstruction learning,”IEEE Trans. Pattern Anal. Mach. Intell., vol. 47, no. 9, pp. 8040–8056, 2025

    work page 2025

  16. [16]

    Cycle-based fre- quency disentanglement diffusion model with self-training for cross- domain hyperspectral-rgb change detection,

    J. Qu, J. Ren, W. Dong, S. Xiao, and Y . Li, “Cycle-based fre- quency disentanglement diffusion model with self-training for cross- domain hyperspectral-rgb change detection,”IEEE Trans. Image Pro- cess., vol. 34, pp. 8130–8144, 2025

    work page 2025

  17. [17]

    Iterative robust graph for unsupervised change detection of heterogeneous remote sensing images,

    Y . Sun, L. Lei, D. Guan, and G. Kuang, “Iterative robust graph for unsupervised change detection of heterogeneous remote sensing images,”IEEE Trans. Image Process., vol. 30, pp. 6277–6291, 2021

    work page 2021

  18. [18]

    Hetecd: Feature consistency alignment and difference mining for heterogeneous remote sensing image change detection,

    W. Jing, H. Bai, B. Song, W. Ni, J. Wu, and Q. Wang, “Hetecd: Feature consistency alignment and difference mining for heterogeneous remote sensing image change detection,”ISPRS J. Photogramm. Remote Sens., vol. 223, pp. 317–327, 2025

    work page 2025

  19. [19]

    Domain adaptive cross reconstruction for change detection of heterogeneous remote sensing images via a feedback guidance mechanism,

    Q. Liu, K. Ren, X. Meng, and F. Shao, “Domain adaptive cross reconstruction for change detection of heterogeneous remote sensing images via a feedback guidance mechanism,”IEEE Trans. Geosci. Remote Sens., vol. 61, pp. 1–16, 2023

    work page 2023

  20. [20]

    Object-based land cover classification and change analysis in the baltimore metropolitan area using multitem- poral high resolution remote sensing data,

    W. Zhou, A. Troy, and M. Grove, “Object-based land cover classification and change analysis in the baltimore metropolitan area using multitem- poral high resolution remote sensing data,”Sensors, vol. 8, no. 3, pp. 1613–1636, 2008

    work page 2008

  21. [21]

    Change alignment-based graph structure learning for unsupervised heterogeneous change detection,

    K. Xiao, Y . Sun, G. Kuang, and L. Lei, “Change alignment-based graph structure learning for unsupervised heterogeneous change detection,” IEEE Geosci. Remote Sens. Lett., vol. 20, pp. 1–5, 2023

    work page 2023

  22. [22]

    Nonlocal patch similarity based heterogeneous remote sensing change detection,

    Y . Sun, L. Lei, X. Li, H. Sun, and G. Kuang, “Nonlocal patch similarity based heterogeneous remote sensing change detection,”Pattern Recogn., vol. 109, p. 107598, 2021

    work page 2021

  23. [23]

    Heterogeneous image change detection based on two-stage joint feature learning,

    T. Han, Y . Tang, and Y . Chen, “Heterogeneous image change detection based on two-stage joint feature learning,” inProceedings of the IEEE International Geoscience and Remote Sensing Symposium, 2022, pp. 3215–3218

    work page 2022

  24. [24]

    A feature space constraint-based method for change detection in heterogeneous images,

    N. Shi, K. Chen, G. Zhou, and X. Sun, “A feature space constraint-based method for change detection in heterogeneous images,”Remote Sens., vol. 12, p. 3057, 09 2020

    work page 2020

  25. [25]

    Change detection with cross-domain remote sensing images: A systematic review,

    J. Chen, D. Hou, C. He, Y . Liu, Y . Guo, and B. Yang, “Change detection with cross-domain remote sensing images: A systematic review,”IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., vol. 17, pp. 11 563–11 582, 2024

    work page 2024

  26. [26]

    Deep learning in remote sensing applications: A meta-analysis and review,

    L. Ma, Y . Liu, X. Zhang, Y . Ye, G. Yin, and B. A. Johnson, “Deep learning in remote sensing applications: A meta-analysis and review,” ISPRS J. Photogramm. Remote Sens., vol. 152, pp. 166–177, 2019

    work page 2019

  27. [27]

    Dual- branch feature fusion network based cross-modal enhanced cnn and transformer for hyperspectral and lidar classification,

    W. Wang, C. Li, P. Ren, X. Lu, J. Wang, G. Ren, and B. Liu, “Dual- branch feature fusion network based cross-modal enhanced cnn and transformer for hyperspectral and lidar classification,”IEEE Geosci. Remote Sens. Lett., vol. 21, pp. 1–5, 2024

    work page 2024

  28. [28]

    Change de- tection in heterogeneous images based on multiple pseudo-homogeneous image pairs,

    H. Zhuang, J. Guo, M. Hao, S. Du, K. Zhang, and X. Wang, “Change de- tection in heterogeneous images based on multiple pseudo-homogeneous image pairs,”Int. J. Appl. Earth Obs. Geoinf., vol. 136, p. 104321, 2025

    work page 2025

  29. [29]

    Change detection in heterogeneous optical and sar remote sensing images via deep homogeneous feature fusion,

    X. Jiang, G. Li, Y . Liu, X.-P. Zhang, and Y . He, “Change detection in heterogeneous optical and sar remote sensing images via deep homogeneous feature fusion,”IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., vol. 13, pp. 1551–1566, 2020

    work page 2020

  30. [30]

    Sar-to-optical image translation based on improved cgan,

    X. Yang, J. Zhao, Z. Wei, N. Wang, and X. Gao, “Sar-to-optical image translation based on improved cgan,”Pattern Recogn., vol. 121, p. 108208, 2022

    work page 2022

  31. [31]

    Unpaired image-to-image translation using cycle-consistent adversarial networks,

    J.-Y . Zhu, T. Park, P. Isola, and A. A. Efros, “Unpaired image-to-image translation using cycle-consistent adversarial networks,” inProceedings of 2017 IEEE/CVF International Conference on Computer Vision, 2017, pp. 2242–2251

    work page 2017

  32. [32]

    A deep translation (gan) based change detection network for optical and sar remote sensing images,

    X. Li, Z. Du, Y . Huang, and Z. Tan, “A deep translation (gan) based change detection network for optical and sar remote sensing images,” ISPRS J. Photogramm. Remote Sens., vol. 179, pp. 14–34, 2021

    work page 2021

  33. [33]

    Unsupervised change detection from heterogeneous data based on image translation,

    Z.-G. Liu, Z.-W. Zhang, Q. Pan, and L.-B. Ning, “Unsupervised change detection from heterogeneous data based on image translation,”IEEE Trans. Geosci. Remote Sens., vol. 60, pp. 1–13, 2022

    work page 2022

  34. [34]

    Afs- net: attention-guided full-scale feature aggregation network for high- resolution remote sensing image change detection,

    M. Jiang, X. Zhang, Y . Sun, W. Feng, Q. Gan, and Y . Ruan, “Afs- net: attention-guided full-scale feature aggregation network for high- resolution remote sensing image change detection,”GISci. Remote Sens., vol. 59, no. 1, pp. 1882–1900, 2022

    work page 1900

  35. [35]

    Multiscale attention network guided with change gradient image for land cover change detection using remote sensing images,

    Z. Lv, P. Zhong, W. Wang, Z. You, and N. Falco, “Multiscale attention network guided with change gradient image for land cover change detection using remote sensing images,”IEEE Geosci. Remote Sens. Lett., vol. 20, pp. 1–5, 2023

    work page 2023

  36. [36]

    Deep multiscale siamese network with parallel convolutional structure and self-attention for change detection,

    Q. Guo, J. Zhang, S. Zhu, C. Zhong, and Y . Zhang, “Deep multiscale siamese network with parallel convolutional structure and self-attention for change detection,”IEEE Trans. Geosci. Remote Sens., vol. 60, pp. 1–12, 2022

    work page 2022

  37. [37]

    Change detection for high-resolution remote sensing images based on a multi-scale attention siamese network,

    J. Li, S. Zhu, Y . Gao, G. Zhang, and Y . Xu, “Change detection for high-resolution remote sensing images based on a multi-scale attention siamese network,”Remote Sens., vol. 14, no. 14, 2022

    work page 2022

  38. [38]

    Hfa-net: High frequency attention siamese network for building change detection in vhr remote sensing images,

    H. Zheng, M. Gong, T. Liu, F. Jiang, T. Zhan, D. Lu, and M. Zhang, “Hfa-net: High frequency attention siamese network for building change detection in vhr remote sensing images,”Pattern Recogn., vol. 129, p. 108717, 2022

    work page 2022

  39. [39]

    A deeply supervised attention metric-based network and an open aerial image 20 dataset for remote sensing change detection,

    Q. Shi, M. Liu, S. Li, X. Liu, F. Wang, and L. Zhang, “A deeply supervised attention metric-based network and an open aerial image 20 dataset for remote sensing change detection,”IEEE Trans. Geosci. Remote Sens., vol. 60, pp. 1–16, 2022

    work page 2022

  40. [40]

    Hfnet: Semantic and differential heterogenous fusion network for remote sensing image change detection,

    Y . Han, J. Li, Y . Qu, L. Wang, X. Pan, and X. Huang, “Hfnet: Semantic and differential heterogenous fusion network for remote sensing image change detection,”J. Geovisual. Spat. Anal., vol. 9, no. 1, p. 1, nov 2024

    work page 2024

  41. [41]

    A bayesian meta-learning-based method for few-shot hyperspectral image classification,

    J. Zhang, L. Liu, R. Zhao, and Z. Shi, “A bayesian meta-learning-based method for few-shot hyperspectral image classification,”IEEE Trans. Geosci. Remote Sens., vol. 61, pp. 1–13, 2023

    work page 2023

  42. [42]

    Hi- erarchical attention feature fusion-based network for land cover change detection with homogeneous and heterogeneous remote sensing images,

    Z. Lv, J. Liu, W. Sun, T. Lei, J. A. Benediktsson, and X. Jia, “Hi- erarchical attention feature fusion-based network for land cover change detection with homogeneous and heterogeneous remote sensing images,” IEEE Trans. Geosci. Remote Sens., vol. 61, pp. 1–15, 2023

    work page 2023

  43. [43]

    Transformer-based multimodal change detection with multitask consistency constraints,

    B. Liu, H. Chen, K. Li, and M. Y . Yang, “Transformer-based multimodal change detection with multitask consistency constraints,”Inf. Fusion, vol. 108, p. 102358, Aug. 2024

    work page 2024

  44. [44]

    Gstm-scd: Graph-enhanced spatio-temporal state space model for semantic change detection in multi-temporal remote sensing images,

    X. Liu, C. Dai, L. Ding, Z. Zhang, Y . Li, X. Zuo, M. Li, H. Wang, and Y . Miao, “Gstm-scd: Graph-enhanced spatio-temporal state space model for semantic change detection in multi-temporal remote sensing images,” ISPRS J. Photogramm. Remote Sens., vol. 230, pp. 73–91, 2025

    work page 2025

  45. [45]

    Exploring foundation models in remote sensing image change detection: A comprehensive survey,

    Z. Yu, T. Li, Y . Zhu, and R. Pan, “Exploring foundation models in remote sensing image change detection: A comprehensive survey,” 2024

    work page 2024

  46. [46]

    Adapting segment anything model for change detection in vhr remote sensing images,

    L. Ding, K. Zhu, D. Peng, H. Tang, K. Yang, and L. Bruzzone, “Adapting segment anything model for change detection in vhr remote sensing images,”IEEE Trans. Geosci. Remote Sens., vol. 62, pp. 1–11, 2024

    work page 2024

  47. [47]

    Scd-sam: Adapting segment anything model for semantic change detection in remote sensing imagery,

    L. Mei, Z. Ye, C. Xu, H. Wang, Y . Wang, C. Lei, W. Yang, and Y . Li, “Scd-sam: Adapting segment anything model for semantic change detection in remote sensing imagery,”IEEE Trans. Geosci. Remote Sens., vol. 62, pp. 1–13, 2024

    work page 2024

  48. [48]

    Peftcd: Leveraging vision foundation models with parameter-efficient fine-tuning for remote sensing change detection,

    S. Dong, Y . Hu, L. Wang, G. Chen, and X. Meng, “Peftcd: Leveraging vision foundation models with parameter-efficient fine-tuning for remote sensing change detection,” 2025

    work page 2025

  49. [49]

    Spacenet 6: Multi-sensor all weather mapping dataset,

    J. Shermeyer, D. Hogan, J. Brown, A. Van Etten, N. Weir, F. Paci- fici, R. H ¨ansch, A. Bastidas, S. Soenen, T. Bacastow, and R. Lewis, “Spacenet 6: Multi-sensor all weather mapping dataset,” in2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2020, pp. 768–777

    work page 2020

  50. [50]

    Unsu- pervised image regression for heterogeneous change detection,

    L. T. Luppino, F. M. Bianchi, G. Moser, and S. N. Anfinsen, “Unsu- pervised image regression for heterogeneous change detection,”IEEE Trans. Geosci. Remote Sens., vol. 57, no. 12, pp. 9960–9975, 2019

    work page 2019

  51. [51]

    A fractal projection and markovian segmentation-based approach for multimodal change detection,

    M. Mignotte, “A fractal projection and markovian segmentation-based approach for multimodal change detection,”IEEE Trans. Geosci. Remote Sens., vol. 58, no. 11, pp. 8046–8058, 2020

    work page 2020

  52. [52]

    Wuhan dataset: A high-resolution dataset of spatiotemporal fusion for remote sensing images,

    X. Zhang, L. Xie, S. Li, F. Lei, L. Cao, and X. Li, “Wuhan dataset: A high-resolution dataset of spatiotemporal fusion for remote sensing images,”IEEE Geosci. Remote Sens. Lett., vol. 21, pp. 1–5, 2024

    work page 2024

  53. [53]

    BRIGHT: a globally distributed multimodal building damage assessment dataset with very-high-resolution for all-weather disaster response,

    H. Chen, J. Song, O. Dietrich, C. Broni-Bediako, W. Xuan, J. Wang, X. Shao, Y . Wei, J. Xia, C. Lan, K. Schindler, and N. Yokoya, “BRIGHT: a globally distributed multimodal building damage assessment dataset with very-high-resolution for all-weather disaster response,”Earth Sys- tem Science Data, vol. 17, no. 11, pp. 6217–6253, 2025

    work page 2025

  54. [54]

    Encoder- decoder with atrous separable convolution for semantic image segmenta- tion,

    L.-C. Chen, Y . Zhu, G. Papandreou, F. Schroff, and H. Adam, “Encoder- decoder with atrous separable convolution for semantic image segmenta- tion,” inProceedings of 2018 European Conference on Computer Vision. Cham: Springer International Publishing, 2018, pp. 833–851

    work page 2018

  55. [55]

    Change detection in multisource vhr images via deep siamese convolutional multiple-layers recurrent neural network,

    H. Chen, C. Wu, B. Du, L. Zhang, and L. Wang, “Change detection in multisource vhr images via deep siamese convolutional multiple-layers recurrent neural network,”IEEE Trans. Geosci. Remote Sens., vol. 58, no. 4, pp. 2848–2864, 2020

    work page 2020

  56. [56]

    Learning from multimodal and multitemporal earth observation data for building damage mapping,

    B. Adriano, N. Yokoya, J. Xia, H. Miura, W. Liu, M. Matsuoka, and S. Koshimura, “Learning from multimodal and multitemporal earth observation data for building damage mapping,”ISPRS J. Photogramm. Remote Sens., vol. 175, pp. 132–143, 2021

    work page 2021

  57. [57]

    Building damage assessment for rapid disaster response with a deep object-based semantic change detection framework: From natural disasters to man- made disasters,

    Z. Zheng, Y . Zhong, J. Wang, A. Ma, and L. Zhang, “Building damage assessment for rapid disaster response with a deep object-based semantic change detection framework: From natural disasters to man- made disasters,”Remote Sens. Environ., vol. 265, p. 112636, 2021

    work page 2021

  58. [58]

    Icif-net: Intra-scale cross-interaction and inter-scale feature fusion network for bitemporal remote sensing images change detection,

    Y . Feng, H. Xu, J. Jiang, H. Liu, and J. Zheng, “Icif-net: Intra-scale cross-interaction and inter-scale feature fusion network for bitemporal remote sensing images change detection,”IEEE Trans. Geosci. Remote Sens., vol. 60, pp. 1–13, 2022

    work page 2022

  59. [59]

    Simple multiscale unet for change detection with heterogeneous remote sensing images,

    Z. Lv, H. Huang, L. Gao, J. A. Benediktsson, M. Zhao, and C. Shi, “Simple multiscale unet for change detection with heterogeneous remote sensing images,”IEEE Geosci. Remote Sens. Lett., vol. 19, pp. 1–5, 2022

    work page 2022

  60. [60]

    Dual- tasks siamese transformer framework for building damage assessment,

    H. Chen, E. Nemni, S. Vallecorsa, X. Li, C. Wu, and L. Bromley, “Dual- tasks siamese transformer framework for building damage assessment,” inProceedings of 2022 IEEE International Geoscience and Remote Sensing Symposium, 2022, pp. 1600–1603

    work page 2022

  61. [61]

    Semantic change detection using a hierarchical semantic graph interaction network from high-resolution remote sensing images,

    J. Long, M. Li, X. Wang, and A. Stein, “Semantic change detection using a hierarchical semantic graph interaction network from high-resolution remote sensing images,”ISPRS J. Photogramm. Remote Sens., vol. 211, pp. 318–335, 2024

    work page 2024

  62. [62]

    Refined change detection in heterogeneous low-resolution remote sensing images for disaster emergency response,

    D. Wang, G. Ma, H. Zhang, X. Wang, and Y . Zhang, “Refined change detection in heterogeneous low-resolution remote sensing images for disaster emergency response,”ISPRS J. Photogramm. Remote Sens., vol. 220, pp. 139–155, 2025

    work page 2025

  63. [63]

    Sigma: Siamese mamba network for multi-modal semantic segmentation,

    Z. Wan, P. Zhang, Y . Wang, S. Yong, S. Stepputtis, K. Sycara, and Y . Xie, “Sigma: Siamese mamba network for multi-modal semantic segmentation,” 2025

    work page 2025