Pith. sign in

REVIEW 4 major objections 7 minor 2 cited by

Towards Interactive Global Geolocation Assistant

T0 review · 4 major / 7 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read The paper claims that an interactive multimodal assistant, GaGA, can geolocate street-view images and refine its answers from user clues, beating prior models on the reproduced GWS15k benchmark at country and city level.

desk verdict The interactive geolocation paradigm and the MG-Geo dataset are genuinely new and worth engaging; the headline SOTA claim rests on an unvalidated reproduction of a closed benchmark and should not be taken at face value. read the letter →

arxiv 2412.08907 v3 pith:4EHIYDUC submitted 2024-12-12 cs.CV

Zhiyang Dou , Zipeng Wang , Xumeng Han , Guorong Li , Zhipei Huang , Zhenjun Han This is my paper · ORCID
classification cs.CV
keywords interactivegeolocationmultimodallargelanguagemodelsglobalchain-of-thoughtreasoningstreet-viewimagesMG-GeodatasetGWS15kbenchmarkvisual
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

GaGA reframes global geolocation as an interactive dialogue rather than a one-shot classification. Built on a multimodal large language model, it reads geographic clues from an image, proposes a country, region, city, and coordinates, and then revises its answer when a user supplies corrections or extra observations. To train this behavior, the paper introduces MG-Geo, a five-million-pair dataset with location metadata, human-style geographic clues, and multi-turn question-answer dialogues generated through chain-of-thought prompting. On a self-reproduced version of the closed GWS15k benchmark, GaGA reaches 63.06% country accuracy and 6.28% city accuracy, outperforming the previous state-of-the-art OSV-5M-Baseline by 4.57 and 2.92 percentage points, and it further improves when given guiding questions or answer priors.

What carries the argument

The mechanism that carries the argument is the MG-Geo dataset, a five-million-pair collection structured into three parts: Meta Part (OSV-5M images paired with country, region, city, and coordinates), Clue Part (GeoGuessr-derived geographic clues matched to images by InternVL2), and Dialog Part (multi-turn question-answer dialogues generated by GPT-4V with chain-of-thought deduction and a 25-km-haversine introspection trigger). GaGA trains on this dataset in two stages: projectors are pretrained on Meta Part, then the Llama3-8B language model is finetuned with QLoRA on a 240k mix. The interactivity emerges from the dialogue format itself, letting each user message update the model's location estimate.

What would settle it

Release or obtain the official GWS15k images and evaluate GaGA directly; if its country and city accuracies do not maintain the reported 4.57% and 2.92% margins over OSV-5M-Baseline, the reproduced-benchmark equivalence is refuted. As a cheaper check, run Algorithm 1 independently and verify that the sampled country distribution and city radii match the paper's reported global distribution (Figure 5).

Watch

Extended reading notes

Core claim

GaGA's central claim is that global geolocation should be treated as an interactive, explainable dialogue rather than a single-shot classification or retrieval step. The authors show that a multimodal large language model, pretrained on geographic metadata and finetuned on clue extraction and dialog data, can name the country, region, and city of a street-view image and then revise that answer when a user supplies corrections or priors. On a reconstructed version of the GWS15k benchmark, GaGA reaches 63.06% country accuracy and 6.28% city accuracy, a lead of 4.57 and 2.92 points over the OSV-5M-Baseline, and it raises its own country accuracy to 74.77% when a question–answer prior is provided. The authors attribute this capacity to the MG-Geo dataset, whose Meta, Clue, and Dialog parts turn five million image-text pairs into a training signal that couples world knowledge with interactive reasoning.

Load-bearing premise

The reproduced GWS15k benchmark, sampled by Algorithm 1 from the OSV-5M test set, faithfully represents the official closed-source GWS15k, so that accuracy gains measured on the reproduction transfer to the real benchmark.

Editorial extensions

If this is right

  • GaGA outperforms the previous state of the art on the reproduced GWS15k: 63.06% vs 58.49% country accuracy and 6.28% vs 3.36% city accuracy.
  • When given a guiding question (+Q), GaGA's region and city accuracy rise, and when given a correct question-answer prior (+QA), its country accuracy jumps from 64.89% to 74.77%.
  • GaGA's coordinate predictions reach the best reported Geoscore (3113.0) on the reproduced GWS15k, beating OSV-5M-Baseline and GeoCLIP.
  • On the full 210K-image OSV-5M test set, GaGA maintains its country- and city-level lead and improves coordinate accuracy at 750 km and 2500 km thresholds, showing the reported gains are not confined to the sampled benchmark.
  • Traditional specialized geolocation models still win at 1 km and 25 km precision, so the MLLM approach's main strength is coarse administrative and long-range accuracy, with fine coordinate output as an open problem.

Reading between the lines

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

  • A real-user study, where humans give natural corrections rather than oracle-answer priors, would test whether the interactive loop's gains survive outside the controlled +QA setup in the paper.
  • The clue-matching design suggests a general recipe: a few hundred expert heuristics can be amplified by an MLLM into millions of image-text pairs, which might transfer to other expert domains like bird identification or architectural dating.
  • Because the headline numbers come from a reconstructed GWS15k, the safest comparison for future work is to run all methods on the full OSV-5M test set (as in Table IV) or on the official GWS15k once released.
  • The CoT Introspection step, which triggers only when predictions are off by more than 25 km, could be reused as a general self-correction curriculum for MLLM reasoning tasks beyond geolocation.
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 / 7 minor

Summary. The paper proposes GaGA, an interactive global geolocation assistant built on a multimodal large language model. It introduces a new dataset, MG-Geo, containing five million image-text pairs, and a two-stage training procedure (geographic pretraining followed by interactive fine-tuning with QLoRA). The authors report state-of-the-art accuracy on the GWS15k benchmark, improving country-level accuracy by 4.57% and city-level accuracy by 2.92% over the OSV-5M-Baseline, along with coordinate-prediction results and interactive-dialogue evaluations.

Significance. Interactive geolocation is a plausible and useful paradigm, and the MG-Geo dataset is a substantial resource. The idea of training an MLLM with geographic metadata, clue extraction, and multi-turn dialogue is well-motivated. The internal comparison between GaGA and OSV-5M-Baseline on the same reproduced split is fair and shows consistent improvement. However, the headline claim of state-of-the-art performance on the official GWS15k benchmark is not supported, because the benchmark used is a self-built reproduction whose equivalence to the closed-source GWS15k is not established. This issue affects the abstract, the SOTA claim, and the comparisons to published numbers in Table II.

major comments (4)
  1. [V-A1, Algorithm 1] The claim that GaGA achieves state-of-the-art performance on GWS15k is based entirely on the reproduction described in Algorithm 1, which samples from the OSV-5M test set around city centers with a 5 km radius. The authors state that GWS15k is not open-source (Section II-B), but they provide no evidence that their sampling procedure yields the same image distribution as the official GWS15k. Without such validation (e.g., distribution statistics, overlap with original samples, or release of image IDs for independent assessment), the external validity of the SOTA claim is unverified. At minimum, the abstract and Section V-B should clearly state that the results are on a reproduced GWS15k subset, not the official benchmark.
  2. [Table II] Table II mixes results evaluated on the official GWS15k (ISNs, Translocator, GeoDecoder, PIGEON) with results evaluated on the reproduced subset (GeoCLIP⋆, OSV-5M-Baseline⋆, GaGA⋆). Comparing GaGA to the non-starred methods across different test distributions is invalid and can mislead readers into concluding that GaGA outperforms methods like PIGEON. The paper should either obtain or compute results on a common benchmark for all methods, or separate the tables and explicitly state that cross-benchmark comparisons are not meaningful.
  3. [V-C, V-D] The evaluation of interactive dialogue compares GaGA against GPT-4V (Table V), but GPT-4V is also the teacher model used to generate the Dialog Part of MG-Geo (Section IV-C). This creates a teacher-student confound: GaGA is fine-tuned on GPT-4V outputs and then evaluated against the same model. Additionally, the expert ratings are based on only 50 samples, and the reported Fleiss' Kappa of 0.55 for Fluency indicates moderate, not high, agreement. The claim that 'the ratings provided by the experts demonstrate a high degree of consistency' should be softened, and the confound should be discussed.
  4. [V-A3, Table I] Table I uses hierarchical (HIER) evaluation for models not fine-tuned on MG-Geo, providing them with candidate administrative names at each level, while GaGA and OSV-5M-Baseline use the direct (DIRE) mode. These protocols are not directly comparable, as HIER supplies additional information that makes the prediction task easier. The claim that GaGA 'outperforms StreetCLIP' on GWS15k is therefore not a fair head-to-head comparison unless both models are evaluated under the same protocol. Please make the protocol differences explicit in the discussion or apply the same mode to all methods.
minor comments (7)
  1. [Section IV-C] The hyperparameter description states that 'temperature and GPT-4V's top-p and top-k parameters are set to 1, 1, and NONE'; top-k is typically an integer or None, and the sentence should be clarified.
  2. [References] References [18] and [43] are the same OSV-5M paper and should be merged to avoid duplication.
  3. [Section III-D2] The line begins with '2)GaGA can also output clues', which contains a stray '2)'.
  4. [Section V-D] The dialog evaluation uses only 50 samples; please report confidence intervals for the win/loss rates and clarify how the 50 samples were selected.
  5. [Appendix Table IX] The heading says 'DIALOGUE' but the text has the typo 'dialoge'; also standardize capitalization across the table.
  6. [Section V-E] The sentence 'we report GaGA's performance on and Im2GPS3k' is missing the dataset name before 'and'.
  7. [General] The paper does not state whether the MG-Geo dataset and the reproduced GWS15k will be released; please include a data availability statement to support reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: accuracy is measured on a held-out test split and no training objective is the evaluation metric.

full rationale

The derivation chain is a standard supervised MLLM pipeline: MG-Geo (built from OSV-5M and Google Landmarks V2) is used to pretrain the projector and QLoRA-finetune the LLM; accuracy is then measured on the GWS15k reproduction, which is explicitly sampled from the OSV-5M test set (Section V-A1, Algorithm 1), not from the training data. The evaluation metrics (admin-level accuracy, Haversine thresholds, Geoscore) are not part of the training loss, and the OSV-5M-Baseline comparison in Table I and Table II is run on the same reproduced split, so the relative improvement is not forced by construction. The main caveat is benchmark fidelity: the paper's GWS15k is a self-built reproduction of the closed-source benchmark, and the authors acknowledge this in Section V-F by also reporting results on the full OSV-5M test set. This is an external-validity concern, not circularity, because no equation or fitted parameter reduces the reported accuracy to the training target. The dialog evaluation against GPT-4V is also not definitionally circular: although GPT-4V generated much of the MG-Geo Dialog Part, the pairwise human ratings are a human assessment rather than a criterion defined as agreement with GPT-4V, so outperforming the teacher on human judgments is not a logical consequence of training on that teacher. No load-bearing self-citations or imported uniqueness theorems appear; reference [32] is an external source for the GWS15k sampling approach. Hence no circular step can be exhibited.

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

No fundamentally new physical entities are introduced. The ledger lists two hand-chosen thresholds that shape the dataset and the reproduced test set, plus four domain assumptions about data reliability and benchmark equivalence.

free parameters (2)
  • CoT introspection distance threshold = 25 km
    Hand-chosen in Section IV-C.1: predicted coordinates from GPT-4V farther than 25 km from the true coordinates trigger the introspection rewriting step. This affects training dialogue quality but is not fitted to the final benchmark.
  • GWS15k reproduction sampling radius = 5 km
    Fixed in Algorithm 1 (Section V-A1); controls which OSV-5M test points enter the reproduced test set and may affect all reported accuracies.
assumptions (4)
  • domain assumption OSV-5M image metadata (country, region, city, GPS) is accurate enough for training and evaluation.
    Used throughout; the Meta Part is built by removing only samples with incomplete annotations (Section IV-A).
  • domain assumption The GWS15k reproduction algorithm yields a test set statistically equivalent to the closed-source GWS15k.
    Section V-A1; no external validation possible because GWS15k is not open-source.
  • domain assumption GPT-4V-generated clue and dialogue data (including introspected corrections with ground-truth coordinates) provides valid supervision.
    Section IV-C; the Dialog Part and Clue Part are generated with GPT-4V and only partially human-verified (Section IV-B3).
  • domain assumption The 547-image interactive evaluation set and its GPT-4V-generated QA prompts are representative of realistic user interactions and do not reveal the ground-truth answer in the question.
    Section V-C; questions are said to be related to visible elements without providing visual details, but the +QA condition supplies the correct answer as context.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards Interactive Global Geolocation Assistant." pith.science (2026). https://pith.science/paper/4EHIYDUC

@misc{pith2026241208907,
  author       = {Pith},
  title        = {Pith review of: Towards Interactive Global Geolocation Assistant},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4EHIYDUC}},
  note         = {Machine review of arXiv:2412.08907}
}
read the original abstract

Global geolocation, which seeks to predict the geographical location of images captured anywhere in the world, is one of the most challenging tasks in the field of computer vision. In this paper, we introduce an innovative interactive global geolocation assistant named GaGA, built upon the flourishing large vision-language models (LVLMs). GaGA uncovers geographical clues within images and combines them with the extensive world knowledge embedded in LVLMs to determine the geolocations while also providing justifications and explanations for the prediction results. We further designed a novel interactive geolocation method that surpasses traditional static inference approaches. It allows users to intervene, correct, or provide clues for the predictions, making the model more flexible and practical. The development of GaGA relies on the newly proposed Multi-modal Global Geolocation (MG-Geo) dataset, a comprehensive collection of 5 million high-quality image-text pairs. GaGA achieves state-of-the-art performance on the GWS15k dataset, improving accuracy by 4.57% at the country level and 2.92% at the city level, setting a new benchmark. These advancements represent a significant leap forward in developing highly accurate, interactive geolocation systems with global applicability.

Figures

Figures reproduced from arXiv: 2412.08907 by the authors.

Figure 1
Figure 1. The mainstream Global Geolocation models. GaGA, trained with the [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The architecture of GaGA, comprising a vision encoder [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Illustrations of GaGA’s dialogues in various scenarios. On the left, we demonstrate how GaGA successfully incorporates external knowledge with [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Data Statistics of Geo-Localized Question-Answer Clue Pairs and Clue Generation Pipeline. (a) The pipeline for generating image-clue pairs. (b) Statistical data on clue types, categorized into eight major categories covering a wide range of topics. (c) Country cue pair…
Figure 5
Figure 5. Figure 5: Distribution of GWS15k. For detailed information on the reproduction [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. GeoRanker: Distance-Aware Ranking for Worldwide Image Geolocalization

    cs.CV 2025-05 conditional novelty 7.0 of 10

    A distance-aware ranking framework with a multi-order loss and a new ranking dataset improves worldwide image geolocalization, achieving state-of-the-art on IM2GPS3K and YFCC4K.

  2. HoloGeo: Mitigating Landmark Bias in Geo-localization via Evidence-Driven Reasoning

    cs.CV 2026-07 reject novelty 6.0 of 10

    A VLM geo-localizer trained with multi-cue rewards improves accuracy on a new landmark-bias benchmark, but the benchmark and training data suffer from unresolved leakage and overlap concerns.

Reference graph

Works this paper leans on

55 extracted references · 29 canonical work pages · cited by 2 Pith papers

  1. [1]

    A survey of location prediction on twitter,

    X. Zheng, J. Han, and A. Sun, “A survey of location prediction on twitter,” IEEE Transactions on Knowledge and Data Engineering , vol. 30, no. 9, pp. 1652–1671, 2018

    work page 2018

  2. [2]

    Metageo: A general framework for social user geolocation identification with few- shot learning,

    F. Zhou, X. Qi, K. Zhang, G. Trajcevski, and T. Zhong, “Metageo: A general framework for social user geolocation identification with few- shot learning,” IEEE Transactions on Neural Networks and Learning Systems, vol. 34, no. 11, pp. 8950–8964, 2023

    work page 2023

  3. [3]

    Cross-view image sequence geo- localization,

    X. Zhang, W. Sultani, and S. Wshah, “Cross-view image sequence geo- localization,” in WACV, 2023

    work page 2023

  4. [4]

    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 IEEE Conference on Computer Vision and Pattern Recognition , 2022, pp. 1162–1171

    work page 2022

  5. [5]

    CPlaNet: Enhancing Image Geolocalization by Combinatorial Partitioning of Maps,

    P. H. Seo, T. Weyand, J. Sim, and B. Han, “CPlaNet: Enhancing Image Geolocalization by Combinatorial Partitioning of Maps,” in Proceedings of the European Conference on Computer Vision , 2018, pp. 536–551

    work page 2018

  6. [6]

    Geoclip: Clip- inspired alignment between locations and images for effective worldwide geo-localization,

    V . Vivanco Cepeda, G. K. Nayak, and M. Shah, “Geoclip: Clip- inspired alignment between locations and images for effective worldwide geo-localization,” Advances in Neural Information Processing Systems , vol. 36, 2024

    2024

  7. [7]

    A latent variable model for geographic lexical variation,

    J. Eisenstein, B. O’Connor, N. A. Smith, and E. Xing, “A latent variable model for geographic lexical variation,” in Proceedings of the 2010 conference on empirical methods in natural language processing , 2010, pp. 1277–1287

    work page 2010

  8. [8]

    Hierarchical geographical modeling of user locations from social media posts,

    A. Ahmed, L. Hong, and A. J. Smola, “Hierarchical geographical modeling of user locations from social media posts,” in Proceedings of the 22nd International Conference on World Wide Web , ser. WWW ’13. Association for Computing Machinery, 2013, p. 25–36

    work page 2013

Show all 55 references
  1. [9]

    Text-based twitter user geolocation prediction,

    B. Han, P. Cook, and T. Baldwin, “Text-based twitter user geolocation prediction,” J. Artif. Int. Res. , vol. 49, no. 1, p. 451–500, Jan. 2014

    2014

  2. [10]

    A neural model for user geolocation and lexical dialectology,

    A. Rahimi, T. Cohn, and T. Baldwin, “A neural model for user geolocation and lexical dialectology,” arXiv preprint arXiv:1704.04008, 2017

    2017 arXiv

  3. [11]

    Kernel density estimation for text-based geolocation,

    M. Hulden, M. Silfverberg, and J. Francom, “Kernel density estimation for text-based geolocation,” in Proceedings of the AAAI conference on artificial intelligence, vol. 29, no. 1, 2015

    2015

  4. [12]

    Find me if you can: im- proving geographical prediction with social and spatial proximity,

    L. Backstrom, E. Sun, and C. Marlow, “Find me if you can: im- proving geographical prediction with social and spatial proximity,” in Proceedings of the 10th International Conference on World Wide Web , ser. WWW ’10. New York, NY , USA: Association for Computing Machinery, 2010...

  5. [13]

    Spot: Locating social media users based on social network context,

    L. Kong, Z. Liu, and Y . Huang, “Spot: Locating social media users based on social network context,”Proceedings of the VLDB Endowment, vol. 7, no. 13, pp. 1681–1684, 2014

    2014

  6. [14]

    Twitter user geolocation us- ing a unified text and network prediction model,

    A. Rahimi, T. Cohn, and T. Baldwin, “Twitter user geolocation us- ing a unified text and network prediction model,” arXiv preprint arXiv:1506.08259, 2015

    2015 arXiv

  7. [15]

    Visual instruction tuning,

    H. Liu, C. Li, Q. Wu, and Y . J. Lee, “Visual instruction tuning,” 2023. [Online]. Available: https://arxiv.org/abs/2304.08485

    2023 arXiv

  8. [16]

    Internvl: Scaling up vision foundation models and aligning for generic visual-linguistic tasks,

    Z. Chen, J. Wu, W. Wang, W. Su, G. Chen, S. Xing, M. Zhong, Q. Zhang, X. Zhu, L. Lu, B. Li, P. Luo, T. Lu, Y . Qiao, and J. Dai, “Internvl: Scaling up vision foundation models and aligning for generic visual-linguistic tasks,” arXiv preprint arXiv:2312.14238 , 2023

    2023 arXiv

  9. [17]

    Kosmos-2: Grounding multimodal large language models to the world,

    Z. Peng, W. Wang, L. Dong, Y . Hao, S. Huang, S. Ma, and F. Wei, “Kosmos-2: Grounding multimodal large language models to the world,” 2023. [Online]. Available: https://arxiv.org/abs/2306.14824

    2023 arXiv

  10. [18]

    Openstreetview-5m: The many roads to global visual geolocation,

    G. Astruc, N. Dufour, I. Siglidis, C. Aronssohn, N. Bouia, S. Fu, R. Loiseau, V . N. Nguyen, C. Raude, E. Vincent, L. Xu, H. Zhou, and L. Landrieu, “Openstreetview-5m: The many roads to global visual geolocation,” in Proceedings of the IEEE/CVF Conference on Computer Vision an...

    2024

  11. [19]

    Sharegpt4v: Improving large multi-modal models with better captions,

    L. Chen, J. Li, X. Dong, P. Zhang, C. He, J. Wang, F. Zhao, and D. Lin, “Sharegpt4v: Improving large multi-modal models with better captions,” 2023. [Online]. Available: https://arxiv.org/abs/2311.12793

    2023 arXiv

  12. [20]

    DataComp: In search of the next generation of multimodal datasets,

    S. Y . Gadre, G. Ilharco, A. Fang, J. Hayase, G. Smyrnis, T. Nguyen, R. Marten, M. Wortsman, D. Ghosh, J. Zhang, E. Orgad, R. Entezari, G. Daras, S. Pratt, V . Ramanujan, Y . Bitton, K. Marathe, S. Mussmann, R. Vencu, M. Cherti, R. Krishna, P. W. Koh, O. Saukh, A. Ratner, S. S...

    2023

  13. [21]

    Mimic-it: Multi-modal in-context instruction tuning,

    B. Li, Y . Zhang, L. Chen, J. Wang, F. Pu, J. Yang, C. Li, and Z. Liu, “Mimic-it: Multi-modal in-context instruction tuning,” 2023. [Online]. Available: https://arxiv.org/abs/2306.05425

    2023 arXiv

  14. [22]

    Qlora: Efficient finetuning of quantized llms,

    T. Dettmers, A. Pagnoni, A. Holtzman, and L. Zettlemoyer, “Qlora: Efficient finetuning of quantized llms,” Advances in Neural Information Processing Systems, vol. 36, 2024

    2024

  15. [23]

    Translocator: local realignment and global remapping enabling accurate translocation detec- tion using single-molecule sequencing long reads,

    Y . Wu, R. Luo, T.-W. Lam, H.-F. Ting, and J. Wang, “Translocator: local realignment and global remapping enabling accurate translocation detec- tion using single-molecule sequencing long reads,” in Proceedings of the 11th ACM International Conference on Bioinformatics, Comput...

    2020

  16. [24]

    Learning transferable visual models from natural language supervision,

    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 models from natural language supervision,” in International conference on machine learning . PMLR, 2021, pp. 8748–8763

    2021

  17. [25]

    Pigeon: Predicting image geolocations,

    L. Haas, M. Skreta, S. Alberti, and C. Finn, “Pigeon: Predicting image geolocations,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 12 893–12 902

    2024

  18. [26]

    Gˆ 3: Ge- olocation via guidebook grounding,

    G. Luo, G. Biamby, T. Darrell, D. Fried, and A. Rohrbach, “Gˆ 3: Ge- olocation via guidebook grounding,” arXiv preprint arXiv:2211.15521 , 2022

    2022 arXiv

  19. [27]

    Learning generalized zero- shot learners for open-domain image geolocalization,

    L. Haas, S. Alberti, and M. Skreta, “Learning generalized zero- shot learners for open-domain image geolocalization,” arXiv preprint arXiv:2302.00275, 2023

    2023 arXiv

  20. [28]

    Georeasoner: Geo-localization with reasoning in street views using a large vision-language model,

    L. Li, Y . Ye, B. Jiang, and W. Zeng, “Georeasoner: Geo-localization with reasoning in street views using a large vision-language model,”

  21. [29]

    Im2gps: estimating geographic information from a single image,

    J. Hays and A. A. Efros, “Im2gps: estimating geographic information from a single image,” in 2008 ieee conference on computer vision and pattern recognition. IEEE, 2008, pp. 1–8

    2008

  22. [30]

    Revisiting im2gps in the deep learning era,

    N. V o, N. Jacobs, and J. Hays, “Revisiting im2gps in the deep learning era,” in Proceedings of the IEEE international conference on computer vision, 2017, pp. 2621–2630. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 14

    2017

  23. [31]

    Interpretable semantic photo geolocation,

    J. Theiner, E. Müller-Budack, and R. Ewerth, “Interpretable semantic photo geolocation,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , 2022, pp. 750–760

    2022

  24. [32]

    Where we are and what we’re looking at: Query based worldwide image geo-localization using hierarchies and scenes,

    B. Clark, A. Kerrigan, P. P. Kulkarni, V . V . Cepeda, and M. Shah, “Where we are and what we’re looking at: Query based worldwide image geo-localization using hierarchies and scenes,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, ...

    2023

  25. [33]

    Google landmarks dataset v2-a large-scale benchmark for instance-level recognition and retrieval,

    T. Weyand, A. Araujo, B. Cao, and J. Sim, “Google landmarks dataset v2-a large-scale benchmark for instance-level recognition and retrieval,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 2575–2584

    2020

  26. [34]

    Qwen-vl: A frontier large vision-language model with versatile abilities,

    J. Bai, S. Bai, S. Yang, S. Wang, S. Tan, P. Wang, J. Lin, C. Zhou, and J. Zhou, “Qwen-vl: A frontier large vision-language model with versatile abilities,” arXiv preprint arXiv:2308.12966 , 2023

    2023 arXiv

  27. [35]

    How far are we to gpt-4v? closing the gap to commercial multimodal models with open-source suites,

    Z. Chen, W. Wang, H. Tian, S. Ye, Z. Gao, E. Cui, W. Tong, K. Hu, J. Luo, Z. Ma et al. , “How far are we to gpt-4v? closing the gap to commercial multimodal models with open-source suites,” arXiv preprint arXiv:2404.16821, 2024

    2024 arXiv

  28. [36]

    Llava-med: Training a large language-and-vision assistant for biomedicine in one day,

    C. Li, C. Wong, S. Zhang, N. Usuyama, H. Liu, J. Yang, T. Naumann, H. Poon, and J. Gao, “Llava-med: Training a large language-and-vision assistant for biomedicine in one day,” 2023. [Online]. Available: https://arxiv.org/abs/2306.00890

    2023 arXiv

  29. [37]

    Geochat: Grounded large vision-language model for remote sensing,

    K. Kuckreja, M. S. Danish, M. Naseer, A. Das, S. Khan, and F. S. Khan, “Geochat: Grounded large vision-language model for remote sensing,” 2023. [Online]. Available: https://arxiv.org/abs/2311.15826

    2023 arXiv

  30. [38]

    Vila-m3: Enhancing vision-language models with medical expert knowledge,

    V . Nath, W. Li, D. Yang, A. Myronenko, M. Zheng, Y . Lu, Z. Liu, H. Yin, Y . M. Law, Y . Tang, P. Guo, C. Zhao, Z. Xu, Y . He, G. Heinrich, S. Aylward, M. Edgar, M. Zephyr, P. Molchanov, B. Turkbey, H. Roth, and D. Xu, “Vila-m3: Enhancing vision-language models with medical e...

    2024 arXiv

  31. [39]

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

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, J. Uszkoreit, and N. Houlsby, “An image is worth 16x16 words: Transformers for image recognition at scale,” 2021. [Online]. Available: https://arxi...

    2021 arXiv

  32. [40]

    Xtuner: A toolkit for efficiently fine-tuning llm,

    X. Contributors, “Xtuner: A toolkit for efficiently fine-tuning llm,” https: //github.com/InternLM/xtuner, 2023

    2023

  33. [41]

    Lmdeploy: A toolkit for compressing, deploying, and serving llm,

    L. Contributors, “Lmdeploy: A toolkit for compressing, deploying, and serving llm,” https://github.com/InternLM/lmdeploy, 2023

    2023

  34. [42]

    Chain-of-thought prompting elicits reasoning in large language models,

    J. Wei, X. Wang, D. Schuurmans, M. Bosma, B. Ichter, F. Xia, E. Chi, Q. Le, and D. Zhou, “Chain-of-thought prompting elicits reasoning in large language models,” 2023. [Online]. Available: https://arxiv.org/abs/2201.11903

    2023 arXiv

  35. [43]

    Openstreetview- 5m: The many roads to global visual geolocation,

    G. Astruc, N. Dufour, I. Siglidis, C. Aronssohn, N. Bouia, S. Fu, R. Loiseau, V . N. Nguyen, C. Raude, E. Vincentet al., “Openstreetview- 5m: The many roads to global visual geolocation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 20...

    2024

  36. [44]

    Language models as zero-shot planners: Extracting actionable knowledge for embodied agents,

    W. Huang, P. Abbeel, D. Pathak, and I. Mordatch, “Language models as zero-shot planners: Extracting actionable knowledge for embodied agents,” in International conference on machine learning. PMLR, 2022, pp. 9118–9147

    2022

  37. [45]

    Swe-bench: Can language models resolve real-world github issues?, 2024,

    C. E. Jimenez, J. Yang, A. Wettig, S. Yao, K. Pei, O. Press, and K. Narasimhan, “Swe-bench: Can language models resolve real-world github issues?, 2024,” URL https://arxiv. org/abs/2310.06770, 2023

    2024 arXiv

  38. [46]

    Vipergpt: Visual inference via python execution for reasoning,

    D. Surís, S. Menon, and C. V ondrick, “Vipergpt: Visual inference via python execution for reasoning,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 11 888–11 898

    2023

  39. [47]

    V oyager: An open-ended embodied agent with large language models,

    G. Wang, Y . Xie, Y . Jiang, A. Mandlekar, C. Xiao, Y . Zhu, L. Fan, and A. Anandkumar, “V oyager: An open-ended embodied agent with large language models,” arXiv preprint arXiv:2305.16291 , 2023

    2023 arXiv

  40. [48]

    Llama 3 model card,

    AI@Meta, “Llama 3 model card,” None, 2024. [Online]. Available: https://github.com/meta-llama/llama3/blob/main/MODEL_CARD.md

    2024

  41. [49]

    Geolocation estima- tion of photos using a hierarchical model and scene classification,

    E. Muller-Budack, K. Pustu-Iren, and R. Ewerth, “Geolocation estima- tion of photos using a hierarchical model and scene classification,” in Proceedings of the European conference on computer vision (ECCV) , 2018, pp. 563–579

    2018

  42. [50]

    Geodecoder: Empowering mul- timodal map understanding,

    F. Qi, M. Dai, Z. Zheng, and C. Wang, “Geodecoder: Empowering mul- timodal map understanding,” arXiv preprint arXiv:2401.15118 , 2024

    2024 arXiv

  43. [51]

    Time-llm: Time series forecasting by reprogramming large language models,

    M. Jin, S. Wang, L. Ma, Z. Chu, J. Y . Zhang, X. Shi, P.-Y . Chen, Y . Liang, Y .-F. Li, S. Pan, and Q. Wen, “Time-llm: Time series forecasting by reprogramming large language models,” 2024. [Online]. Available: https://arxiv.org/abs/2310.01728

    2024 arXiv

  44. [52]

    Better and faster large language models via multi-token prediction,

    F. Gloeckle, B. Y . Idrissi, B. Rozière, D. Lopez-Paz, and G. Synnaeve, “Better and faster large language models via multi-token prediction,”

  45. [53]

    Measuring nominal scale agreement among many raters,

    J. L. Fleiss, “Measuring nominal scale agreement among many raters,” Psychological Bulletin, vol. 76, no. 5, pp. 378–382, 1971. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 15 BIOGRAPHY SECTION Zhiyang Dou received the B.E. degree in communication engineering from...

    1971

  46. [54]

    Available: https://arxiv.org/abs/2404.19737

    [Online]. Available: https://arxiv.org/abs/2404.19737

    arXiv

  47. [2024]

    Available: https://arxiv.org/abs/2406.18572

    [Online]. Available: https://arxiv.org/abs/2406.18572

Pith tools

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