Pith. sign in

REVIEW 3 major objections 5 minor 43 references

Reliability-weighted token fusion makes UAV–satellite geo-localization substantially more robust to realistic image corruption, the paper argues, backed by an 11.7-million-image degraded benchmark.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 02:09 UTC pith:N4I3U3XH

load-bearing objection A genuinely useful robustness benchmark and a fusion method that appears to work, with the main caveat being how much of the comparison rests on under-specified baseline conditions. the 3 major comments →

arxiv 2607.25524 v1 pith:N4I3U3XH submitted 2026-07-28 cs.CV cs.AI

ReLATE: Reliability-Guided Evidence Fusion for Robust UAV--Satellite cross-view Geo-Localization

Haochen Jiang , Jialei Pan , Yuzhe Sun , Zhe Dong , Lecheng Ren , Yanfeng Gu , Tianzhu Liu This is my paper
classification cs.CV cs.AI
keywords UAV-satellite geo-localizationcross-view retrievalcorruption robustnessreliability-guided fusiontoken reliabilityclean-training corrupted-testingremote sensing benchmarkfeature fusion
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper claims that UAV–satellite geo-localization, though accurate on clean benchmarks, degrades sharply under real-world visual corruptions such as blur, noise, weather, and compression. To expose this, it builds UAVSat-Deg, a clean-training corrupted-testing benchmark with 27 corruption types at three severity levels across two datasets and two retrieval directions. To fix the problem, it introduces ReLATE, a reliability-guided evidence fusion framework that estimates a spatially smoothed per-token reliability field, suppresses unreliable token responses, and adaptively injects aggregated reliable local evidence into the final descriptor. The paper reports that ReLATE achieves the best average corrupted-test performance on both datasets and both retrieval directions while maintaining or slightly improving clean accuracy, with gains that widen as severity increases.

Core claim

The central claim is that explicitly modeling which local visual evidence can be trusted, and regulating how much that evidence contributes to the retrieval descriptor, makes cross-view geo-localization robust to unseen degradations without any corruption labels or degradation-specific training. ReLATE's SRE module learns a token-wise reliability score, smooths it with local averaging to enforce spatial consistency, and modulates token responses so that above-average-reliability tokens are enhanced and below-average tokens suppressed. The RATE module then aggregates reliable token evidence with concentration-controlled softmax weights and injects it into query-derived representations with an

What carries the argument

The load-bearing mechanism is the structure-smoothed reliability field èRv learned end-to-end: a lightweight MLP with sigmoid produces per-token reliability scores, a learnable α-blend mixes raw scores with 3×3 average pooling to enforce spatial continuity, and mean-centered modulation factors rescale token responses. RATE converts the same field into evidence aggregation weights via a concentration-controlled softmax and computes a regulation coefficient λ from the reliability distribution's mean and standard deviation, then adds λe to remapped query branches. The final descriptor concatenates the regulated query branches, the CLS token, and a GeM-pooled spatial branch, trained jointly with

Load-bearing premise

The headline comparison assumes every baseline was trained and evaluated under the same backbone and protocol as ReLATE, but the paper does not state the backbones, training recipes, or checkpoint sources for the compared methods, so the reported margins could partly reflect encoder or training differences rather than reliability guidance itself.

What would settle it

Train ReLATE and the strongest baseline from the same visual foundation backbone with identical epochs, data augmentation, batch size, and loss weighting, then evaluate on University-1652-Deg; if ReLATE's All-27 corrupted average advantage over the baseline shrinks to near zero, the claimed reliability-guidance benefit would be unsupported.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • If the central claim holds, reliability-guided fusion is a general, corruption-agnostic robustness mechanism that requires no knowledge of the degradation type or severity at test time.
  • The gains are systematic rather than cherry-picked: ReLATE ranks first in 39 of 54 evaluation conditions on each dataset and never falls below third place across all 108 conditions.
  • The benefit grows with degradation strength, suggesting that reliability gating matters most exactly when local visual evidence is least trustworthy.
  • The approach transfers across retrieval directions and UAV heights, improving both query-side robustness (Drone→Satellite) and gallery-side robustness (Satellite→Drone) on multi-height SUES-200-Deg.
  • UAVSat-Deg itself provides a fixed, reproducible benchmark of 11,762,010 pre-generated corrupted test images, enabling controlled future comparisons under an image-only clean-training corrupted-testing protocol.

Where Pith is reading between the lines

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

  • The learned reliability field could plausibly serve as a calibration signal for when a retrieval system should be distrusted under novel degradations, since it is trained without corruption labels and appears to track structural saliency such as building contours and road layouts.
  • Because SRE and RATE are defined on top of a general token representation, the mechanism may transfer to other retrieval substrates or backbones beyond the query-driven substrate used here, though the paper does not demonstrate this.
  • The benchmark currently corrupts only the UAV side; extending it to satellite-side perturbations or real captured degradations, as the paper mentions as future work, could expose different robustness failure modes that pure synthetic corruption does not reveal.
  • A natural testable extension is combining reliability gating with test-time adaptation: the reliability field could prioritize which tokens to trust when adapting a clean-trained model to a newly encountered corruption type.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper introduces UAVSat-Deg, a robustness benchmark for UAV–satellite cross-view geo-localization that applies 27 corruption types at three severity levels to UAV-side test images from University-1652 and SUES-200, under a clean-training / corrupted-testing protocol with bidirectional retrieval and multi-height settings. It reports a total of 11,762,010 pre-generated corrupted test images. The paper then proposes ReLATE, a reliability-guided feature-fusion framework built on a query-based cross-view substrate. ReLATE estimates a structure-smoothed token reliability field (SRE), uses it to modulate spatial tokens and refine query representations, aggregates reliability-weighted token evidence, and adaptively injects this evidence into query-derived descriptor branches (RATE). Experiments on University-1652-Deg and SUES-200-Deg report that ReLATE achieves the best average corrupted-test performance among the compared methods while maintaining clean accuracy, with ablations attributing the gains to both SRE and RATE.

Significance. If the comparisons are fair, the paper makes a useful contribution: it provides a standardized, corruption-aware evaluation protocol for an under-tested setting, and a method whose reliability-guided fusion yields consistent gains over a strong query-based baseline—e.g., +4.36 R@1 on University-1652-Deg D2S All-27 (Table 10). Strengths include the fixed offline corruption generation shared by all methods, the explicit clean-training/corrupted-testing protocol, the broad corruption taxonomy, and mechanism-level ablations (α=0, Global-λ) that help isolate the contributions of structural smoothing and input-dependent regulation. The principal risks are comparison fairness and reproducibility: the manuscript does not specify the backbone, training recipe, or checkpoint source for most baselines, and the benchmark dataset and code are not yet released. Both are central to the paper's claims and need to be addressed before the results can be fully assessed.

major comments (3)
  1. [§5.2–§5.3, Tables 5–8] The baseline comparison is the load-bearing part of the paper's central claim, but the training conditions of the baselines are not specified. Only ReLATE's DINOv2-B/14 setup and hyperparameters are given. Table 3 labels the backbone as DINOv2-B/14, yet it is not stated that all methods in Tables 5–8 use this backbone or the same training recipe. Table 10's 'Base' row exactly reproduces QDFL's clean and All-27 numbers in Table 5 (95.00/95.83 and 65.39/68.55), strongly suggesting that Base is a DINOv2 reimplementation of QDFL, but this is never stated. For DAC, CAMP, Sample4Geo, MCCG, MEAN, MuSe-Net, CCR, and FSRA, no backbone, input size, loss, or checkpoint source is given. If these baselines used weaker or differently trained encoders, the reported margins—and the attribution of gains to SRE/RATE rather than to encoder strength—would not be established. Please provide a complete baseli
  2. [§3.1, §5.2] The benchmark construction is not sufficiently reproducible from the text. Weather corruptions are said to follow a 'WeatherPrompt-style synthesis procedure' [14], but the concrete generation parameters are not given; the remaining corruptions are described only as 'procedural image operators following the common-corruption paradigm' with unspecified presets. Because UAVSat-Deg is itself a core contribution, exact operators, parameter values, severity schedules, and generation code should be provided. The statement that 'code and dataset will be available' is not enough to verify the 11.7M-image benchmark or to enable future comparisons; please release the dataset and code at least in a form accessible to reviewers.
  3. [§5.2, Tables 5–9] All reported numbers appear to be single-run point estimates; no error bars, standard deviations, or number of seeds are given. Some margins in the central comparison are small—e.g., Table 9, SUES-200-Deg S2D at H200 shows Ours at −0.20/−0.96 relative to QDFL. Given that the paper's main claim is comparative ('best average corrupted-test performance'), the absence of uncertainty quantification makes it difficult to know which cell-level differences are meaningful. Please report mean±std over at least three training runs, or explicitly state that only one seed was used and interpret small margins accordingly.
minor comments (5)
  1. [Eq. (1)] The SNR formula uses the uncentered signal energy E[I^2] in the numerator, so it is a distortion measure rather than a physical SNR. Please clarify this and consider reporting the full distribution (e.g., percentiles), not only the median 9.05 dB.
  2. [Eq. (11)] The softmax temperature τ is introduced as a positive concentration controller but its initialization, whether it is learned, and its value in the experiments are not reported. Please specify.
  3. [Table 4] The rank-based rating methodology is described only in the table caption. Please move a concise explanation of the scoring into the main text, including how ties are handled.
  4. [Figure 2] The figure contains the typo 'UA VSat-Deg' for UAVSat-Deg. Please correct the spacing/lettering.
  5. [§3.1] The text says 'The generated files are organized by dataset, direction, family, corruption type, and severity,' but no file naming or directory layout is given. This is a minor reproducibility aid and should be documented once the dataset is released.

Circularity Check

0 steps flagged

No significant circularity: benchmark protocol and ablations keep prediction and fitting separate.

full rationale

ReLATE is an empirical architecture paper; there is no formal derivation chain whose conclusion is built into its premises. The reliability estimator is trained end-to-end on clean images via the retrieval objective ('ReLATE does not rely on any explicit reliability annotation or degradation label. The reliability field is learned implicitly, driven entirely by the retrieval objective,' Sec. 4.4) and then frozen at test time on corrupted images generated from the test splits. The corrupted-test numbers therefore measure generalization to held-out degradations rather than a fitted quantity recycled as a prediction. The controlled ablation (Table 10) compares Base, Base+SRE, Base+RATE†, and Full; the Base row reproduces QDFL's clean and All-27 numbers exactly, so the QDFL comparison is implicitly backbone-controlled, and the neutral RATE† variant and the α=0 / Global-λ mechanism ablations isolate the new modules without reusing the final result as an input. The benchmark itself is a fixed offline corruption cache with clean-training/corrupted-testing constraints; no method is trained or selected on degraded images (Sec. 3.1), so the evaluation cannot be circular by construction. The main uncertainty noted by the reviewer is the absence of stated training/backbone details for DAC, CAMP, etc. in Sec. 5.2. That is a comparison-fairness and reproducibility risk, not a circularity: the compared numbers are independently generated evaluations, and even if the backbone were uncontrolled, the claim would be empirically confounded rather than definitionally forced. Several related-work citations involve co-author Y. Sun, but they are background/baseline references (e.g., MuSe-Net [12]) and are not used to justify any load-bearing step of ReLATE's derivation. Hence no circular step is present.

Axiom & Free-Parameter Ledger

6 free parameters · 5 axioms · 0 invented entities

No new physical entities are postulated; the 'reliability field' is a learned internal representation, not an independently testable entity. The central claims rest on benchmark-protocol assumptions and on several hand-chosen hyperparameters, all listed above.

free parameters (6)
  • softmax temperature tau in Eq. (11)
    Controls concentration of token evidence weights; not reported as learned or as a chosen value, and it is load-bearing for RATE aggregation.
  • structural smoothing coefficient alpha (Eq. 6)
    Learned through sigmoid parameterization, but initialization and final value are not reported; it controls how much local neighborhood smoothing is applied to the reliability field.
  • modulation strength gamma (Eq. 8)
    Learned via softplus parameterization; initialization and final value are not reported; it controls the enhance/suppress magnitude applied to tokens.
  • architecture sizes Kq and Nsup = Kq=8, Nsup=2
    Hand-chosen capacities that determine the number of internal queries and supervised descriptor branches; no sensitivity study is provided.
  • Multi-Similarity loss constants = alpha_MS=5, beta_MS=100, b_MS=0.1, epsilon_MS=0.4
    Adopted from prior metric learning work but still chosen constants that shape the embedding space; they are not derived in this paper.
  • equal loss weights in L = L_cls + L_MS + L_cv = 1, 1, 1
    The paper states no extra loss-balancing hyperparameters are introduced; equal weighting is a simplification that affects the training trade-off.
axioms (5)
  • domain assumption Synthetic corruptions generated from WeatherPrompt/common-corruption procedures are representative of real UAV degradations and preserve location-discriminative content.
    Section 3.1: the benchmark relies on this to make corrupted-test rankings meaningful; the authors list real-captured degradations as future work.
  • domain assumption Only UAV-side test images are corrupted; satellite images remain clean.
    Section 3.1: motivated by moving onboard cameras, but real satellite imagery can also be degraded by atmosphere or compression; this limits the generality of the conclusions.
  • domain assumption A token-level reliability field learned purely from clean retrieval supervision generalizes to unseen corruption types and severities.
    Core hypothesis of ReLATE in Section 4.2: no reliability labels or corruption labels are used, so generalization is assumed and only indirectly supported by held-out test results.
  • standard math L2-normalized inner product (Eq. 3) is a valid retrieval criterion for the learned descriptors.
    Standard retrieval setup inherited from prior work; the paper does not prove any metric property of the final descriptor.
  • domain assumption DINOv2-B/14 features provide a sufficient visual substrate for cross-view matching under degradation.
    Section 4.1: the method is instantiated with only this backbone; if foundation features are themselves non-robust, ReLATE may not compensate; no other backbone is tested.

pith-pipeline@v1.3.0-alltime-deepseek · 32088 in / 15215 out tokens · 154305 ms · 2026-08-01T02:09:55.671728+00:00 · methodology

0 comments
read the original abstract

Unmanned aerial vehicle (UAV)-satellite cross-view geo-localization matches UAV images against satellite imagery and has achieved impressive accuracy on clean (non-degraded) image benchmarks. In real-world flights, however, UAV observations are frequently affected by adverse weather, illumination changes, platform motion, sensor noise, and compression, while the robustness of existing methods under such degradations remains largely unexamined. In this paper, we present UAVSat-Deg, a large-scale robustness benchmark for degraded UAV-satellite geo-localization, comprising University-1652-Deg and SUES-200-Deg. UAVSat-Deg covers 27 corruption types, including 19 core and 8 compound corruptions, at three severity levels, supports bidirectional drone-to-satellite and satellite-to-drone retrieval as well as multi-height UAV acquisition, and contains more than 11.7 million pre-generated corrupted test images. Benchmarking representative methods under this protocol reveals substantial robustness gaps, particularly under severe and compound corruptions. To address this problem, we propose ReLATE, a Reliable Evidence Learning framework with Adaptive Token Evidence Regulation, which realizes reliability-adaptive feature fusion during descriptor construction. ReLATE estimates a structure-smoothed reliability field over visual tokens, aggregates trustworthy local evidence, and adaptively integrates it into query-derived representations; the regulated query representations are then combined with the CLS-token and GeM-pooled branches to form the final cross-view descriptor. Across both test sets and retrieval directions, ReLATE achieves the best average corrupted-test performance among the compared methods while maintaining competitive accuracy on clean images. The code and dataset will be available at https://github.com/JHC626/ReLATE.

Figures

Figures reproduced from arXiv: 2607.25524 by Haochen Jiang, Jialei Pan, Lecheng Ren, Tianzhu Liu, Yanfeng Gu, Yuzhe Sun, Zhe Dong.

Figure 1
Figure 1. Figure 1: Motivation of reliability-guided evidence aggregation [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the UAVSat-Deg corruption taxonomy. The benchmark contains 27 corruption types at severity levels 1, 2, and 3: [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Overall framework of the proposed ReLATE. UAV-view and satellite-view images are independently encoded by a shared image [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Illustration of structure-smoothed reliability-guided evi [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Visualization of the structure-smoothed reliability field [PITH_FULL_IMAGE:figures/full_fig_p010_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Qualitative retrieval results on University-1652-Deg under representative degraded UAV queries. Each row corresponds to one [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Visualization of the structure-smoothed reliability field [PITH_FULL_IMAGE:figures/full_fig_p014_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Qualitative retrieval results on SUES-200-Deg under representative degraded UAV queries. Each row corresponds to one degraded [PITH_FULL_IMAGE:figures/full_fig_p015_8.png] view at source ↗

discussion (0)

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

Reference graph

Works this paper leans on

43 extracted references · 3 linked inside Pith

  1. [1]

    Zheng, Y

    Z. Zheng, Y. Wei, Y. Yang, University-1652: A multi- view multi-source benchmark for drone-based geo- localization, in: Proceedings of the 28th ACM inter- national conference on Multimedia, 2020, pp. 1395– 1403

    2020

  2. [2]

    R. Zhu, L. Yin, M. Yang, F. Wu, Y. Yang, W. Hu, Sues-200: A multi-height multi-scene cross-view im- age benchmark across drone and satellite, IEEE Transactions on Circuits and Systems for Video Tech- nology 33 (2023) 4825–4839. 18

    2023

  3. [3]

    T.-Y. Lin, Y. Cui, S. Belongie, J. Hays, Learning deep representations for ground-to-aerial geolocaliza- tion, in: Proceedings of the IEEE conference on com- puter vision and pattern recognition, 2015, pp. 5007– 5015

    2015

  4. [4]

    S. Hu, M. Feng, R. M. Nguyen, G. H. Lee, Cvm- net: Cross-view matching network for image-based ground-to-aerial geo-localization, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 7258–7267

    2018

  5. [5]

    Y. Shi, L. Liu, X. Yu, H. Li, Spatial-aware feature ag- gregation for image based cross-view geo-localization, Advances in Neural Information Processing Systems 32 (2019)

    2019

  6. [6]

    H. Yang, X. Lu, Y. Zhu, Cross-view geo-localization with layer-to-layer transformer, Advances in Neu- ral Information Processing Systems 34 (2021) 29009– 29020

    2021

  7. [7]

    S. Zhu, M. Shah, C. Chen, Transgeo: Transformer is all you need for cross-view image geo-localization, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 1162–1171

    2022

  8. [8]

    Deuser, K

    F. Deuser, K. Habel, N. Oswald, Sample4geo: Hard negative sampling for cross-view geo-localisation, in: Proceedings of the IEEE/CVF International Confer- ence on Computer Vision, 2023, pp. 16847–16856

    2023

  9. [9]

    Dosovitskiy, L

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weis- senborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, et al., An im- age is worth 16x16 words: Transformers for image recognition at scale, arXiv preprint arXiv:2010.11929 (2020)

    Pith/arXiv arXiv 2010

  10. [10]

    Oquab, T

    M. Oquab, T. Darcet, T. Moutakanni, H. Vo, M. Szafraniec, V. Khalidov, P. Fernandez, D. Haz- iza, F. Massa, A. El-Nouby, et al., Dinov2: Learn- ing robust visual features without supervision, arXiv preprint arXiv:2304.07193 (2023)

    Pith/arXiv arXiv 2023

  11. [11]

    Zhang, Y

    Q. Zhang, Y. Zhu, Benchmarking the robustness of cross-view geo-localization models, in: European Conference on Computer Vision, Springer, 2024, pp. 36–53

    2024

  12. [12]

    T. Wang, Z. Zheng, Y. Sun, C. Yan, Y. Yang, T.- S. Chua, Multiple-environment self-adaptive network for aerial-view geo-localization, Pattern Recognition 152 (2024) 110363

    2024

  13. [13]

    T. Feng, Q. Li, X. Wang, M. Wang, G. Li, W. Zhu, Multi-weather cross-view geo-localization using de- noising diffusion models, in: Proceedings of the 2nd Workshop on UAVs in Multimedia: Capturing the World from a New Perspective, 2024, pp. 35–39

    2024

  14. [14]

    J. Wen, H. Yu, Z. Zheng, Weatherprompt: Multi- modalityrepresentationlearningforall-weatherdrone visual geo-localization, Advances in Neural Informa- tion Processing Systems 38 (2026) 32974–32994

    2026

  15. [15]

    Workman, R

    S. Workman, R. Souvenir, N. Jacobs, Wide-area im- age geolocalization with aerial reference imagery, in: Proceedings of the IEEE International Conference on Computer Vision, 2015, pp. 3961–3969

    2015

  16. [16]

    T. Wang, Z. Zheng, C. Yan, J. Zhang, Y. Sun, B.Zheng, Y.Yang, Eachpartmatters: Localpatterns facilitate cross-view geo-localization, IEEE Transac- tions on Circuits and Systems for Video Technology 32 (2021) 867–879

    2021

  17. [17]

    M. Dai, J. Hu, J. Zhuang, E. Zheng, A transformer- based feature segmentation and region alignment method for uav-view geo-localization, IEEE Transac- tions on Circuits and Systems for Video Technology 32 (2021) 4376–4389

    2021

  18. [18]

    L. Ding, J. Zhou, L. Meng, Z. Long, A practical cross- view image matching method between uav and satel- lite for uav-based geo-localization, Remote Sensing 13 (2020) 47

    2020

  19. [19]

    X. Tian, J. Shao, D. Ouyang, H. T. Shen, Uav- satellite view synthesis for cross-view geo-localization, IEEE Transactions on Circuits and Systems for Video Technology 32 (2021) 4804–4815

    2021

  20. [20]

    J. Lin, Z. Zheng, Z. Zhong, Z. Luo, S. Li, Y. Yang, N. Sebe, Joint representation learning and keypoint detection for cross-view geo-localization, IEEE Trans- actions on Image Processing 31 (2022) 3780–3792

    2022

  21. [21]

    T. Shen, Y. Wei, L. Kang, S. Wan, Y.-H. Yang, Mccg: Aconvnext-basedmultiple-classifiermethodforcross- view geo-localization, IEEE Transactions on Circuits and Systems for Video Technology 34 (2023) 1456– 1468

    2023

  22. [22]

    P. Xia, Y. Wan, Z. Zheng, Y. Zhang, J. Deng, Enhancing cross-view geo-localization with domain alignment and scene consistency, IEEE Transactions on Circuits and Systems for Video Technology 34 (2024) 13271–13281

    2024

  23. [23]

    Q. Wu, Y. Wan, Z. Zheng, Y. Zhang, G. Wang, Z. Zhao, Camp: A cross-view geo-localization method using contrastive attributes mining and position- aware partitioning, IEEE Transactions on Geoscience and Remote Sensing 62 (2024) 1–14

    2024

  24. [24]

    J. Lin, Z. Luo, D. Lin, S. Li, Z. Zhong, A self- adaptive feature extraction method for aerial-view geo-localization, IEEE Transactions on Image Pro- cessing 34 (2024) 126–139. 19

    2024

  25. [25]

    H. Du, J. He, Y. Zhao, Ccr: A counterfactual causal reasoning-based method for cross-view geo- localization, IEEE Transactions on Circuits and Sys- tems for Video Technology 34 (2024) 11630–11643

    2024

  26. [26]

    S. Hu, Z. Shi, T. Jin, Y. Liu, Query-driven feature learning for cross-view geo-localization, IEEE Trans- actions on Geoscience and Remote Sensing (2025)

    2025

  27. [27]

    Rodrigues, M

    R. Rodrigues, M. Tani, Global assists local: Ef- fective aerial representations for field of view con- strained image geo-localization, in: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2022, pp. 3871–3879

    2022

  28. [28]

    B. Sun, G. Liu, Y. Yuan, F3-net: Multiview scene matching for drone-based geo-localization, IEEE Transactions on Geoscience and Remote Sensing 61 (2023) 1–11

    2023

  29. [29]

    T. Wang, Z. Zheng, Z. Zhu, Y. Sun, C. Yan, Y. Yang, Learning cross-view geo-localization embeddings via dynamic weighted decorrelation regularization, IEEE Transactions on Geoscience and Remote Sensing 62 (2024) 1–12

    2024

  30. [30]

    Q. Chen, T. Wang, Z. Yang, H. Li, R. Lu, Y. Sun, B. Zheng, C. Yan, Sdpl: Shifting-dense partition learning for uav-view geo-localization, IEEE Transac- tions on Circuits and Systems for Video Technology 34 (2024) 11810–11824

    2024

  31. [31]

    H. Zhao, K. Ren, T. Yue, C. Zhang, S. Yuan, Transfg: A cross-view geo-localization of satellite and uavs im- agery pipeline using transformer-based feature aggre- gation and gradient guidance, IEEE Transactions on Geoscience and Remote Sensing 62 (2024) 1–12

    2024

  32. [32]

    H. Ju, S. Huang, S. Liu, Z. Zheng, Video2bev: Trans- forming drone videos to bevs for video-based geo- localization, in: Proceedings of the IEEE/CVF Inter- national Conference on Computer Vision, 2025, pp. 27073–27083

    2025

  33. [33]

    M. Chu, Z. Zheng, W. Ji, T. Wang, T.-S. Chua, To- wards natural language-guided drones: Geotext-1652 benchmark with spatial relation matching, in: Euro- pean Conference on Computer Vision, Springer, 2024, pp. 213–231

    2024

  34. [34]

    Q. Zhao, J. Zhou, T. Wang, Q. Chen, R. Lu, C. Yan, P2fcn: Environment-independent uav-view geo-localization via pixel-to-feature co-enhancement, IEEE Transactions on Geoscience and Remote Sens- ing 63 (2025) 1–12

    2025

  35. [35]

    Hendrycks, T

    D. Hendrycks, T. Dietterich, Benchmarking neural network robustness to common corruptions and per- turbations, arXiv preprint arXiv:1903.12261 (2019)

    Pith/arXiv arXiv 1903

  36. [36]

    Radford, J

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, et al., Learning transferable visual mod- els from natural language supervision, in: Interna- tional conference on machine learning, PmLR, 2021, pp. 8748–8763

    2021

  37. [37]

    K. He, X. Zhang, S. Ren, J. Sun, Deep residual learn- ingforimagerecognition, in: ProceedingsoftheIEEE conference on computer vision and pattern recogni- tion, 2016, pp. 770–778

    2016

  38. [38]

    Huang, Z

    G. Huang, Z. Liu, L. Van Der Maaten, K. Q. Wein- berger, Densely connected convolutional networks, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 4700–4708

    2017

  39. [39]

    X.Pan, P.Luo, J.Shi, X.Tang, Twoatonce: Enhanc- ing learning and generalization capacities via ibn-net, in: Proceedings of the european conference on com- puter vision (ECCV), 2018, pp. 464–479

    2018

  40. [40]

    Z. Liu, Y. Lin, Y. Cao, H. Hu, Y. Wei, Z. Zhang, S. Lin, B. Guo, Swin transformer: Hierarchical vision transformer using shifted windows, in: Proceedings of the IEEE/CVF international conference on computer vision, 2021, pp. 10012–10022

    2021

  41. [41]

    Vaswani, N

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, I. Polosukhin, At- tention is all you need, Advances in neural informa- tion processing systems 30 (2017)

    2017

  42. [42]

    X. Wang, X. Han, W. Huang, D. Dong, M. R. Scott, Multi-similarity loss with general pair weight- ing for deep metric learning, in: Proceedings of the IEEE/CVF conference on computer vision and pat- tern recognition, 2019, pp. 5022–5030

    2019

  43. [43]

    Chen, Z.-X

    Z. Chen, Z.-X. Yang, H.-J. Rong, Multi-level em- bedding and alignment network with consistency and invariance learning for cross-view geo-localization, IEEE Transactions on Geoscience and Remote Sens- ing (2025). 20

    2025