Pith. sign in

REVIEW 4 major objections 4 minor 49 references

CV-Cities: Advancing Cross-View Geo-Localization in Global Cities

T0 review · 4 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a DINOv2-based shared-weight framework with a feature mixer, symmetric InfoNCE loss, and GPS-guided hard-negative sampling achieves the highest cross-view geo-localization accuracy to date, including 82.91% average…

desk verdict A genuinely useful new dataset, but the paper's headline SOTA claim on it depends on undocumented baseline training. read the letter →

arxiv 2411.12431 v1 pith:OD7RQBCM submitted 2024-11-19 cs.CV

Gaoshuang Huang , Yang Zhou , Luying Zhao , Wenjian Gan This is my paper · ORCID
classification cs.CV
keywords cross-viewgeo-localizationglobalbenchmarkdatasetDINOv2featuremixinghardnegativesamplingsymmetricInfoNCElosssatelliteimageretrievalvisualplacerecognition
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

Cross-view geo-localization is the task of finding where a ground photo was taken by matching it against satellite imagery, and this paper argues that the field's accuracy limit is set less by the matching architecture than by the data and by how negatives are chosen. The authors propose a shared-weight framework that starts from the self-supervised DINOv2 vision backbone, aggregates its features with a lightweight feature mixer, trains with symmetric InfoNCE loss, and mines hard negatives first by GPS proximity and then by learned similarity. On their new CV-Cities dataset, which holds 223,736 ground-satellite pairs from sixteen cities on six continents, the framework reports 82.91% average top-1 accuracy on six held-out cities, 8.42 points above the previous best. If the comparisons are fair, this is evidence that large pretrained visual features plus hard-negative mining can localize a ground image to the correct city block across very different climates, seasons, and urban geometries, without GNSS.

What carries the argument

The load-bearing mechanism is the feature-extraction-and-retrieval pipeline. A DINOv2 backbone (a self-supervised Vision Transformer; the paper uses the base and small variants) is stripped of its final layer normalization and head, and its $N \times D$ token matrix is reshaped into $s$ feature maps of size $h \times w$. These maps are aggregated by a feature mixer: a stack of $L$ isotropic MLP blocks, each computing $X_i \leftarrow W_2 \sigma(W_1 X_i) + X_i$, followed by depth and line projections and L2 normalization to produce the global descriptor. Training uses the symmetric InfoNCE loss, and the two negative-mining strategies (near-neighbor sampling by GPS distance, then dynamic similarity sampling by cosine similarity) select the hard examples that drive the learned metric. Shared weights between the ground and satellite branches remove the need for polar-coordinate conversion or GAN-based view synthesis.

What would settle it

Re-train Sample4Geo (and at least one other baseline such as SAIG-D) on the identical ten CV-Cities training cities using the paper's own augmentation and evaluation protocol, then measure top-1 on the six test cities; if the re-trained baseline matches or exceeds 82.91%, the claimed superiority reduces to a comparison artifact rather than a property of the proposed framework.

Watch

Extended reading notes

Core claim

The central discovery is that a single recipe transfers across all major CVGL benchmarks: use DINOv2's ViTb14 as a backbone, replace the final norm and head with a feature-transformation layer into feature maps, pass those through two residual MLP feature-mixer blocks, and train with the symmetric InfoNCE loss while sampling negatives in two stages. During the first stage, negatives are chosen by geographic distance (haversine for WGS84 coordinates, Euclidean for UTM) because nearby places look alike; during the second, the model itself re-ranks all samples by cosine similarity and packs the hardest ones into the batch. On CVUSA, CVACT, University-1652, VIGOR, and the new CV-Cities, the resulting ViTb14-mix model reports the highest top-1, AP, or hit rate in every task, including 97.43% top-1 on University-1652 Drone2Sat and a VIGOR CROSS hit rate of 75.97% versus 69.87% for Sample4Geo. The paper also reports that its ViTs14-mix variant, with 22.7 million parameters, is competitive, and that the CV-Cities test set, unlike CVUSA or CVACT, is not saturated.

Load-bearing premise

The reported state-of-the-art numbers on CV-Cities rest on the unstated assumption that every baseline in Table III was retrained on the same ten training cities with comparable hyperparameters and the same evaluation code, and the paper does not describe how those baseline numbers were obtained.

Editorial extensions

If this is right

  • With the reported 82.91% average top-1 on CV-Cities held-out cities, a model trained only on ten cities can localize street-level queries in unseen cities on other continents with high accuracy, suggesting global coverage is not required for global operation.
  • The improvement from random sampling (32.13% top-1) to NNS+DSS (82.91%) implies that negative-sample choice is the dominant training factor, not just the backbone.
  • Symmetric InfoNCE's 15.89- to 17.08-point advantage over triplet losses means the loss function itself is a major accuracy lever.
  • The 22.7-million-parameter ViTs14-mix variant remains competitive, so the recipe scales down to deployment-friendly models.
  • The saturation of CVUSA (99.19% top-1) and CVACT Val (92.59%) observed here supports the paper's case that new benchmarks with held-out cities are needed to measure progress.

Reading between the lines

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

  • A fair community rerun of the Table III baselines on the same ten-city CV-Cities training split would be the decisive test of whether the 8.42-point margin comes from the architecture or from unequally tuned comparisons; the paper does not describe the baseline training protocol.
  • Because ground panoramas are randomly rotated during training, the learned descriptor should be orientation-invariant; an untested corollary is that it may survive perspective (limited-field-of-view) queries better than orientation-aligned models, which the authors identify as future work.
  • If GPS-tagged negatives are the fuel for the first training stage, then in datasets without coordinates one could synthesize nearby-looking views from the satellite imagery itself, extending the recipe beyond the six benchmarks.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The manuscript proposes a cross-view geo-localization (CVGL) framework built on DINOv2 with an MLP-mixer feature aggregation module, symmetric InfoNCE loss, and two negative-sampling strategies (near-neighbor sampling and dynamic similarity sampling). It also introduces CV-Cities, a global dataset of 223,736 ground-satellite pairs across sixteen cities, with six cities held out for testing. The paper reports state-of-the-art results on CVUSA, CVACT, University-1652, VIGOR, and CV-Cities, plus ablations of backbone, mixer depth, parameter updating, loss, and sampling. The abstract and Section IV state that code and data are publicly released.

Significance. If the empirical claims are verified, the main contributions are a large global benchmark and a strong retrieval pipeline. The paper has several genuinely positive features: evaluation on multiple external benchmarks in addition to the new dataset, held-out test cities with no overlap with training, a broad ablation study, and a stated public release of code and data. The dataset could be a useful community resource, and the framework is a credible strong baseline. However, the headline claim rests on cross-method comparisons for which the training protocol on the new dataset is not reported, and the manuscript nowhere reports variance. These are fixable reporting issues rather than mathematical flaws, but they currently limit confidence in the stated margins.

major comments (4)
  1. [Section IV-C, Table III] The CV-Cities column contains numbers for methods such as LPN, SAFA, TransGeo, GeoDTR, SAIG-D, and Sample4Geo, but CV-Cities is introduced in this paper, so those entries cannot come from earlier publications. The manuscript does not state which ten cities were used to train each baseline, which hyperparameters, epochs, batch sizes, augmentations, or negative-sampling settings were applied, or whether the same retrieval and evaluation code was used for all methods. The headline 8.42-point gain over Sample4Geo (82.91 vs 74.49) is therefore not auditable. Please provide full per-baseline configurations and verification scripts, or clearly separate numbers reproduced with the released code from numbers copied from prior papers.
  2. [Section IV-C, Tables III-V] Every reported number appears to come from a single training run, with no error bars or significance tests. Some comparisons are too small to interpret without variance information, notably the SAME-mode top-1 difference of 0.41 points in Table V (ViTb14-mix 78.27 vs Sample4Geo 77.86). Please run at least three seeds per configuration and report mean and standard deviation; if that is infeasible, say so explicitly and identify which differences are reliable.
  3. [Sections II-B, II-C, III-A, Table I] The dataset description contains inconsistencies that affect usability. CVACT is described as Canberra in Section II-B but as Melbourne in Section II-C; the continent count is six in the abstract and Section III-A but five in Section II-B; Table I lists CVUSA training satellite images as 335,532 while the text says 35,532, and lists CVACT training pairs as 44,416 while the text says 35,532; and Section III-A's statement that 'the CV-Cities dataset has eight cities (sixteen in total)' is unclear. Please correct these inconsistencies and add a complete city list with train/test partition in the dataset documentation.
  4. [Sections III-C and III-D] The proposed sampling scheme and loss are not specified to a reproducible level. Equation (7) gives only one direction of InfoNCE, not the symmetric objective that is claimed; the manuscript does not define the schedule for switching from near-neighbor sampling to dynamic similarity sampling, nor the meaning of the 'preliminary training phase'; and the interaction of the sampled negatives with the batch of size 32 is unclear. Since the ablations show that these components change top-1 accuracy by tens of points (Table IX), precise pseudocode or explicit pointers to the released code are needed to verify the central method.
minor comments (4)
  1. [Equation (1)] The notation in Equation (1) is confusing: it states N = hw and D = s, but the surrounding text uses D for the feature length and s for the number of feature maps; please rename one of these to avoid conflating the descriptor dimension with the feature-map count.
  2. [Section IV-B] The mention of label smoothing of 0.1 'within' the InfoNCE loss is not formalized; the exact objective with label smoothing should be written explicitly.
  3. [Section IV-E2] The statement that localization is successful when the distance to the true location is less than 10 m should be reconciled with the top-1 exact-match metric used elsewhere; the color classes in Fig. 10 include 10-500 m as a category, so the threshold definition needs clarification.
  4. [Fig. 7] The caption of Fig. 7 contains a typo: 'mix nodule' should be 'mix module'.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the framework's accuracy is evaluated on external benchmarks and held-out cities, with no reduction between inputs and predictions.

full rationale

The paper's derivation chain is not circular. The method (DINOv2 backbone, feature-mix module from MLP-Mixer, symmetric InfoNCE loss, NNS/DSS sampling) is assembled from externally published components, and the central performance claim is measured on held-out test cities of CV-Cities plus independent public benchmarks (CVUSA, CVACT, University-1652, VIGOR). No equation in Sections III-B through III-D defines the reported accuracy in terms of the method's own outputs; top-k, AP, and hit-rate are standard metrics applied to held-out image pairs. The ablations in Tables VI-IX tune design choices such as mix-layer number, parameter-update combinations, and sampling parameters on the CV-Cities test split, which is a model-selection-on-test-set concern rather than a constructional circularity. The CV-Cities baseline column in Table III is not accompanied by a description of how each baseline was retrained or evaluated; this is an important reproducibility gap, but it does not make the comparison circular by construction. The only self-citation is reference [5], a background survey cited in the introduction for CVGL challenges; it is not load-bearing, and no uniqueness theorem or prior result by the same authors is invoked to force the framework choice. The limitations acknowledged in Section V (panoramic-image dependence, water/cloud failure cases, and geographic coverage) describe scope restrictions, not circular steps. Because no prediction reduces by definition to a fitted input or to a self-citation chain, the paper receives score 0.

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

The central claim is empirical. It rests on the validity of the new dataset and on a set of hyperparameters (mixer depth, negative-sample sizes, update strategy) chosen by experiments on the CV-Cities test set. No new physical or conceptual entities are introduced.

free parameters (4)
  • Number of feature mixer layers (L) = 2
    Ablated over 1 to 4 on CV-Cities (Fig. 7); L=2 chosen for final results, tuned on the test set.
  • Negative sampling sizes (S, s) = S=128, s=64
    Set in Section III.C.2 with no ablation; directly controls the DSS hard-negative mining and affects accuracy.
  • Parameter update strategy = All blocks + mix
    Table VII shows the best update set differs by dataset; 'All' is used for CV-Cities main results and was selected based on CV-Cities test accuracy.
  • InfoNCE temperature (tau) = not reported
    Temperature in Eq. (7) is not specified, although it is a sensitive hyperparameter in contrastive losses.
assumptions (3)
  • domain assumption DINOv2 pretrained weights provide transferable visual features for cross-view matching.
    The method relies on DINOv2's pretrained representations; if these do not transfer, the model's performance drops. Section III.B.2.
  • domain assumption Each ground-satellite image pair in CV-Cities is correctly aligned and the one-to-one retrieval target is well-defined.
    Dataset construction (Section III.A) assumes the satellite image centered at the ground sample point is the unique best match, despite 100m spacing and 222m satellite footprints that overlap. If neighboring satellites are equally good matches, top-1 accuracy becomes an unreliable metric.
  • standard math Standard InfoNCE loss derivation.
    Symmetric InfoNCE is taken from prior work (Sample4Geo, CLIP) without proof. It is a standard contrastive loss, not a new derivation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CV-Cities: Advancing Cross-View Geo-Localization in Global Cities." pith.science (2026). https://pith.science/paper/OD7RQBCM

@misc{pith2026241112431,
  author       = {Pith},
  title        = {Pith review of: CV-Cities: Advancing Cross-View Geo-Localization in Global Cities},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OD7RQBCM}},
  note         = {Machine review of arXiv:2411.12431}
}
read the original abstract

Cross-view geo-localization (CVGL), which involves matching and retrieving satellite images to determine the geographic location of a ground image, is crucial in GNSS-constrained scenarios. However, this task faces significant challenges due to substantial viewpoint discrepancies, the complexity of localization scenarios, and the need for global localization. To address these issues, we propose a novel CVGL framework that integrates the vision foundational model DINOv2 with an advanced feature mixer. Our framework introduces the symmetric InfoNCE loss and incorporates near-neighbor sampling and dynamic similarity sampling strategies, significantly enhancing localization accuracy. Experimental results show that our framework surpasses existing methods across multiple public and self-built datasets. To further improve globalscale performance, we have developed CV-Cities, a novel dataset for global CVGL. CV-Cities includes 223,736 ground-satellite image pairs with geolocation data, spanning sixteen cities across six continents and covering a wide range of complex scenarios, providing a challenging benchmark for CVGL. The framework trained with CV-Cities demonstrates high localization accuracy in various test cities, highlighting its strong globalization and generalization capabilities. Our datasets and codes are available at https://github.com/GaoShuang98/CVCities.

Figures

Figures reproduced from arXiv: 2411.12431 by the authors.

Figure 1
Figure 1. The city and sample points distribution map of CV-Cities datasets. (a) The city distribution map of the CV-Cities dataset, with red and green dots [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Examples of ground and satellite images of different types of scenes. The left image of each scene pair is the ground image, and the right image is [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. The distribution of images in CV-Cities on scenes, yearly and monthly scale. (a) Scenes distribution. (b) Yearly distribution. (c) Monthly distribution. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Our framework of CVGL [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: The structures of DINOv2 and feature transformation. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 8
Figure 8. Figure 8: The precision distribution of the ViTb14-mix that training on the [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Example of CVGL in the CV-Cities using our framework. The first column is the query image, and columns two to six are the retrieved top [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: The precision distribution in CV-Cities’ six test cities of our model. [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]
Figure 11
Figure 11. Figure 11: Visualization of heatmaps. on the CV-Cities, which presents more complex scenarios, the accuracy improvement reaches 8.42%. Unlike CVUSA and CVACT, the CV-Cities is larger, more geographically diverse, and features a wider range of scenes, greatly enhancing the model’…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 17 canonical work pages

  1. [1]

    Wide-Area Geolocalization with a Limited Field of View Camera in Challenging Urban Environments

    L. M. Downes, T. J. Steiner, R. L. Russell, and J. P. How, “Wide-Area Geolocalization with a Limited Field of View Camera in Challenging Urban Environments,” arXiv, Aug. 14, 2023. Accessed: Dec. 03, 2023. [Online]. Available: http://arxiv.org/abs/2308.07432

    work page Pith review arXiv 2023

  2. [2]

    Spatial-Aware Feature Aggregation for Cross-View Image based Geo-Localization,

    Y . Shi, L. Liu, X. Yu, and H. Li, “Spatial-Aware Feature Aggregation for Cross-View Image based Geo-Localization,” in Advances in Neural In- formation Processing Systems 32 (nips 2019) , H. Wallach, H. Larochelle, A. Beygelzimer, F. d’Alche-Buc, E. Fox, and R. Garnett, Eds., La Jolla: Neural Information Processing Systems (nips) , 2019. Accessed: Sep. 21,

    work page 2019

  3. [3]

    Ground-to-Aerial Image Geo-Localization with a Hard Exemplar Reweighting Triplet Loss,

    S. Cai, Y . Guo, S. Khan, J. Hu, and G. Wen, “Ground-to-Aerial Image Geo-Localization with a Hard Exemplar Reweighting Triplet Loss,” in 2019 IEEE/CVF International Conference on Computer Vision (ICCV) , New York: Ieee, Oct. 2019, pp. 8390–8399. doi: 10.1109/ICCV .2019.00848

    arXiv 2019

  4. [4]

    Image-Based Geo-Localization Using Satellite Imagery,

    S. Hu and G. H. Lee, “Image-Based Geo-Localization Using Satellite Imagery,” Int. J. Comput. Vis. , vol. 128, no. 5, pp. 1205–1219, May 2020, doi: 10.1007/s11263-019-01186-0

    work page doi:10.1007/s11263-019-01186-0 2020

  5. [5]

    A Survey of the Research Progress in Image Geo- localization,

    G. S. Huang, Y . Zhou, X. F. Hu, L. Y . Zhao, and C. L. Zhang, “A Survey of the Research Progress in Image Geo- localization,” Journal of Geo-information Science , vol. 25, no. 7, pp. 1336–1362, 2023, doi: 10.12082/dqxxkx.2023.230073

    work page arXiv 2023

  6. [6]

    DINOv2: Learning Robust Visual Features without Supervision,

    M. Oquab et al., “DINOv2: Learning Robust Visual Features without Supervision,” arXiv, Apr. 14, 2023. doi: 10.48550/arxiv.2304.07193

    arXiv doi:10.48550/arxiv.2304.07193 2023

  7. [7]

    MLP-Mixer: An all-MLP Architecture for Vision,

    I. O. Tolstikhin et al., “MLP-Mixer: An all-MLP Architecture for Vision,” in Advances in Neural Information Processing Systems , M. Ranzato, A. Beygelzimer, Y . Dauphin, P. S. Liang, and J. W. Vaughan, Eds., Curran Associates, Inc., 2021, pp. 24261–24272. doi: 10.48550/arXiv.2105.01601

    arXiv doi:10.48550/arxiv.2105.01601 2021

  8. [8]

    Sample4Geo: Hard Negative Sampling For Cross-View Geo-Localisation

    F. Deuser, K. Habel, and N. Oswald, “Sample4Geo: Hard Negative Sam- pling For Cross-View Geo-Localisation,”arXiv, Aug. 29, 2023. Accessed: Nov. 16, 2023. [Online]. Available: http://arxiv.org/abs/2303.11851

    work page Pith review arXiv 2023

Show all 49 references
  1. [9]

    A Simple Framework for Contrastive Learning of Visual Representations,

    T. Chen, S. Kornblith, M. Norouzi, and G. Hinton, “A Simple Framework for Contrastive Learning of Visual Representations,” inProceedings of the 37th Interntional Conference on Machine Learning , 2020

    2020

  2. [10]

    Momentum Contrast for Unsupervised Visual Representation Learning,

    K. He, H. Fan, Y . Wu, S. Xie, and R. Girshick, “Momentum Contrast for Unsupervised Visual Representation Learning,” in 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, W A, USA: IEEE, Jun. 2020, pp. 9726–9735. doi: 10.1109/CVPR42600.2020.00975

    2020

  3. [11]

    Cross-View Image Geolocal- ization,

    T.-Y . Lin, S. Belongie, and J. Hays, “Cross-View Image Geolocal- ization,” in 2013 IEEE Conference on Computer Vision and Pattern Recognition, Jun. 2013, pp. 891–898. doi: 10.1109/CVPR.2013.120

    2013 doi

  4. [12]

    Histograms of Oriented Gradients for Human Detection,

    N. Dalal and B. Triggs, “Histograms of Oriented Gradients for Human Detection,” in 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’05) , San Diego, CA, USA: IEEE, 2005, pp. 886–893. doi: 10.1109/CVPR.2005.177

    2005 doi

  5. [13]

    Modeling the Shape of the Scene: A Holistic Representation of the Spatial Envelope,

    A. Oliva and A. Torralba, “Modeling the Shape of the Scene: A Holistic Representation of the Spatial Envelope,” Int. J. Comput. Vis. , vol. 42, no. 3, pp. 145–175, 2001

    2001

  6. [14]

    Vision based robot localization by ground to satellite matching in GPS-denied situations,

    A. Viswanathan, B. R. Pires, and D. Huber, “Vision based robot localization by ground to satellite matching in GPS-denied situations,” in 2014 IEEE/RSJ International Conference on Intelligent Robots and Systems, Chicago, IL, USA: IEEE, Sep. 2014, pp. 192–198. doi: 10.1109/IROS...

    2014

  7. [15]

    Distinctive image features from scale-invariant keypoints,

    D. Lowe, “Distinctive image features from scale-invariant keypoints,” International Journal of Computer Vision , vol. 60, no. 2, pp. 91–110, Nov. 2004, doi: 10.1023/B:VISI.0000029664.99615.94

    2004

  8. [16]

    SURF: Speeded Up Robust Features,

    H. Bay, T. Tuytelaars, and L. Van Gool, “SURF: Speeded Up Robust Features,” in Computer Vision – ECCV 2006 , vol. 3951, A. Leonardis, H. Bischof, and A. Pinz, Eds., in Lecture Notes in Computer Science, vol. 3951. , Berlin, Heidelberg: Springer Berlin Heidelberg, 2006, pp. 404–417

    2006

  9. [17]

    FREAK: Fast Retina Keypoint,

    A. Alahi, R. Ortiz, and P. Vandergheynst, “FREAK: Fast Retina Keypoint,” in 2012 IEEE Conference on Computer Vision and Pat- tern Recognition , Providence, RI: IEEE, Jun. 2012, pp. 510–517. doi: 10.1109/CVPR.2012.6247715

    2012

  10. [18]

    Learning Deep Represen- tations for Ground-to-Aerial Geolocalization,

    T.-Y . Lin, Yin Cui, S. Belongie, and J. Hays, “Learning Deep Represen- tations for Ground-to-Aerial Geolocalization,” in 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , Jun. 2015, pp. 5007–5015. doi: 10.1109/cvpr.2015.7299135

    2015

  11. [19]

    Wide-Area Image Geolo- calization with Aerial Reference Imagery,

    S. Workman, R. Souvenir, and N. Jacobs, “Wide-Area Image Geolo- calization with Aerial Reference Imagery,” in 2015 IEEE International Conference on Computer Vision (ICCV) , Santiago, Chile: IEEE, Dec. 2015, pp. 3961–3969. doi: 10.1109/ICCV .2015.451

    2015 doi

  12. [20]

    Cross-View Image Matching for Geo-localization in Urban Environments,

    Y . Tian, C. Chen, and M. Shah, “Cross-View Image Matching for Geo-localization in Urban Environments,” in 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , New York: IEEE, Jul. 2017, pp. 1998–2006. doi: 10.1109/CVPR.2017.216

    2017 doi

  13. [21]

    Faster R-CNN: Towards Real- Time Object Detection with Region Proposal Networks,

    S. Ren, K. He, R. Girshick, and J. Sun, “Faster R-CNN: Towards Real- Time Object Detection with Region Proposal Networks,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 39, no. 6, pp. 1137–1149, Jun. 2017, doi: 10.1109/TPAMI.2016.2577031

    2017

  14. [22]

    CVM-Net: Cross-View Matching Network for Image-Based Ground-to-Aerial Geo- Localization,

    S. Hu, M. Feng, R. M. H. Nguyen, and G. H. Lee, “CVM-Net: Cross-View Matching Network for Image-Based Ground-to-Aerial Geo- Localization,” in 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition , New York: IEEE, Jun. 2018, pp. 7258–7267. doi: 10.1109/CVPR.2018.00758

    2018

  15. [23]

    NetVLAD: CNN Architecture for Weakly Supervised Place Recognition,

    R. Arandjelovic, P. Gronat, A. Torii, T. Pajdla, and J. Sivic, “NetVLAD: CNN Architecture for Weakly Supervised Place Recognition,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 40, no. 6, pp. 1437–1451, Jun. 2018, doi: 10.1109/TPAMI.2017.2711011

    2018

  16. [24]

    Patch similarity self-knowledge distillation for cross-view geo-localization,

    S. Li, M. Hu, X. Xiao, and Z. Tu, “Patch similarity self-knowledge distillation for cross-view geo-localization,” IEEE Trans. Circuits Syst. Video Technol. , vol. 34, no. 6, pp. 5091–5103, Jun. 2024, doi: 10.1109/TCSVT.2023.3336844

    2024

  17. [25]

    A Transformer-Based Feature Segmentation and Region Alignment Method For UA V-View Geo- Localization,

    M. Dai, J. Hu, J. Zhuang, and E. Zheng, “A Transformer-Based Feature Segmentation and Region Alignment Method For UA V-View Geo- Localization,” IEEE Trans. Circuits Syst. Video Technol. , vol. 32, no. 7, pp. 4376–4389, Jul. 2022, doi: 10.1109/TCSVT.2021.3135013

    2022

  18. [26]

    Cross-view Geo-localization with Evolving Transformer,

    H. Yang, X. Lu, and Y . Zhu, “Cross-view Geo-localization with Evolving Transformer,” arXiv, Jul. 04, 2021. doi: 10.48550/arXiv.2107.00842

    2021 doi arXiv

  19. [27]

    A Semantic Guidance and Transformer-Based Matching Method for UA Vs and Satellite Images for UA V Geo-Localization,

    J. Zhuang, X. Chen, M. Dai, W. Lan, Y . Cai, and E. Zheng, “A Semantic Guidance and Transformer-Based Matching Method for UA Vs and Satellite Images for UA V Geo-Localization,” IEEE Access , vol. 10, pp. 34277–34287, 2022, doi: 10.1109/ACCESS.2022.3162693

    2022

  20. [28]

    Transformer-Guided Convo- lutional Neural Network for Cross-View Geolocalization,

    T. Wang, S. Fan, D. Liu, and C. Sun, “Transformer-Guided Convo- lutional Neural Network for Cross-View Geolocalization,” arXiv, vol. abs/2204.09967, 2022, doi: 10.48550/arXiv.2204.09967

    2022 doi arXiv

  21. [29]

    Cross-View Geo- Localization via Learning Disentangled Geometric Layout Correspon- dence,

    X. Zhang, X. Li, W. Sultani, Y . Zhou, and S. Wshah, “Cross-View Geo- Localization via Learning Disentangled Geometric Layout Correspon- dence,” Proc. AAAI Conf. Artif. Intell. , vol. 37, no. 3, pp. 3480–3488, Jun. 2023, doi: 10.1609/aaai.v37i3.25457

    2023 doi

  22. [30]

    Ground–Satellite Coupling for Cross-View Ge- olocation Combined With Multiscale Fusion of Spatial Features,

    L. Zhao et al., “Ground–Satellite Coupling for Cross-View Ge- olocation Combined With Multiscale Fusion of Spatial Features,” IEEE Geosci. Remote Sens. Lett. , vol. 21, pp. 1–5, 2024, doi: 10.1109/LGRS.2024.3388574. ARXIV 15

    2024

  23. [31]

    Cross-View Image Synthesis Using Conditional GANs,

    K. Regmi and A. Borji, “Cross-View Image Synthesis Using Conditional GANs,” in 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA: IEEE, Jun. 2018, pp. 3501–3510. doi: 10.1109/CVPR.2018.00369

    2018

  24. [32]

    Multi- Channel Attention Selection GAN With Cascaded Semantic Guidance for Cross-View Image Translation,

    H. Tang, D. Xu, N. Sebe, Y . Wang, J. J. Corso, and Y . Yan, “Multi- Channel Attention Selection GAN With Cascaded Semantic Guidance for Cross-View Image Translation,” in 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , Long Beach, CA, USA: IEEE, Ju...

    2019

  25. [34]

    Ground-to-Aerial Image Geo-Localization with Cross-View Image Synthesis,

    J. Huang and D. Ye, “Ground-to-Aerial Image Geo-Localization with Cross-View Image Synthesis,” in Image and Graphics (icig 2021) , Pt Iii, Y . Peng, S. M. Hu, M. Gabbouj, K. Zhou, M. Elad, and K. Xu, Eds., Cham: Springer International Publishing Ag, 2021, pp. 412–424

    2021

  26. [35]

    Lending Orientation to Neural Networks for Cross- View Geo-Localization,

    L. Liu and H. Li, “Lending Orientation to Neural Networks for Cross- View Geo-Localization,” in 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , Long Beach, CA, USA: IEEE, Jun. 2019, pp. 5617–5626. doi: 10.1109/CVPR.2019.00577

    2019

  27. [36]

    University-1652: A Multi-view Multi- source Benchmark for Drone-based Geo-localization,

    Z. Zheng, Y . Wei, and Y . Yang, “University-1652: A Multi-view Multi- source Benchmark for Drone-based Geo-localization,” in Proceedings of the 28th ACM International Conference on Multimedia , Seattle W A USA: ACM, Oct. 2020, pp. 1395–1403. doi: 10.1145/3394171.3413896

    2020

  28. [37]

    VIGOR: Cross-View Image Geo-localization beyond One-to-one Retrieval,

    S. Zhu, T. Yang, and C. Chen, “VIGOR: Cross-View Image Geo-localization beyond One-to-one Retrieval,” in 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , Los Alamitos: Ieee Computer Soc, Jun. 2021, pp. 5316–5325. doi: 10.1109/CVPR46437.2021.00364

    2021

  29. [38]

    Cross-View Visual Geo-Localization for Out- door Augmented Reality,

    N. C. Mithun et al., “Cross-View Visual Geo-Localization for Out- door Augmented Reality,” in 2023 IEEE Conference Virtual Real- ity and 3D User Interfaces (VR) , Mar. 2023, pp. 493–502. doi: 10.1109/vr55154.2023.00064

    2023

  30. [39]

    Each Part Matters: Local Patterns Facilitate Cross-view Geo-localization,

    T. Wang et al., “Each Part Matters: Local Patterns Facilitate Cross-view Geo-localization,” IEEE Trans. Circuits Syst . Video Technol., vol. 32, no. 2, pp. 867–879, Feb. 2022, doi: 10.1109/TCSVT.2021.3061265

    2022

  31. [40]

    TransGeo: Transformer Is All You Need for Cross-view Image Geo-localization,

    S. Zhu, M. Shah, and C. Chen, “TransGeo: Transformer Is All You Need for Cross-view Image Geo-localization,” in 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , New Orleans, LA, USA: IEEE, Jun. 2022, pp. 1152–1161. doi: 10.1109/CVPR52688.2022.00123

    2022

  32. [41]

    Simple, Effective and General: A New Backbone for Cross-view Image Geo-localization,

    Y . Zhu, H. Yang, Y . Lu, and Q. Huang, “Simple, Effective and General: A New Backbone for Cross-view Image Geo-localization,” arXiv, Feb. 03, 2023. Accessed: Nov. 27, 2023. [Online]. Available: http://arxiv.org/abs/2302.01572

    2023 arXiv

  33. [42]

    Google Street View: Capturing the World at Street Level,

    D. Anguelov et al., “Google Street View: Capturing the World at Street Level,” Computer, vol. 43, no. 6, pp. 32–38, Jun. 2010, doi: 10.1109/MC.2010.170

    2010 doi

  34. [43]

    Segment Anything,

    A. Kirillov et al., “Segment Anything,” arXiv, Apr. 05, 2023. Accessed: Oct. 12, 2023. [Online]. Available: http://arxiv.org/abs/2304.02643

    2023 arXiv

  35. [44]

    MixVPR: Feature Mixing for Visual Place Recognition,

    A. Ali-Bey, B. Chaib-Draa, and P. Giguere, “MixVPR: Feature Mixing for Visual Place Recognition,” in 2023 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) , Waikoloa, HI, USA: IEEE, Jan. 2023, pp. 2997–3006. doi: 10.1109/wacv56688.2023.00301

    2023

  36. [45]

    Learning Transferable Visual Models From Natural Language Supervision,

    A. Radford et al., “Learning Transferable Visual Models From Natural Language Supervision,” arXiv. [Online]. Available: http://arxiv.org/abs/2103.00020

    arXiv

  37. [46]

    Image-based Geolocaliza- tion by Ground-to-2.5D Map Matching,

    M. Zhou, L. Liu, Y . Zhong, and A. Calway, “Image-based Geolocaliza- tion by Ground-to-2.5D Map Matching,” arXiv, Nov. 03, 2023. Accessed: Dec. 03, 2023. [Online]. Available: http://arxiv.org/abs/2308.05993

    2023 arXiv

  38. [47]

    Revisiting Street-to-Aerial View Image Geo-localization and Orientation Estimation,

    S. Zhu, T. Yang, and C. Chen, “Revisiting Street-to-Aerial View Image Geo-localization and Orientation Estimation,” in 2021 IEEE Winter Conference on Applications of Computer Vision (WACV) , Los Alamitos: IEEE Computer Soc, Jan. 2021, pp. 756–765. doi: 10.1109/W ACV48630.2021.00080

    2021

  39. [48]

    Multi-scale attention encoder for street-to-aerial image geo-localization,

    S. Li, Z. Tu, Y . Chen, and T. Yu, “Multi-scale attention encoder for street-to-aerial image geo-localization,” CAAI Trans. Intell. Technol. , vol. 8, no. 1, pp. 166–176, Mar. 2023, doi: 10.1049/cit2.12077

    2023 doi

  40. [49]

    ’UA V’s Status Is Worth Considering: A Fusion Representations Matching Method for Geo-Localization,

    R. Zhu, M. Yang, L. Yin, F. Wu, and Y . Yang, “’UA V’s Status Is Worth Considering: A Fusion Representations Matching Method for Geo-Localization,” Sens., vol. 23, no. 2, p. 720, Jan. 2023, doi: 10.3390/s23020720

    2023 doi

  41. [2022]

    Available: https://www.webofscience.com/wos/alldb/full- record/WOS:000535866901069

    [Online]. Available: https://www.webofscience.com/wos/alldb/full- record/WOS:000535866901069

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.