{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,17]],"date-time":"2026-04-17T09:16:44Z","timestamp":1776417404720,"version":"3.51.2"},"reference-count":23,"publisher":"MDPI AG","issue":"4","license":[{"start":{"date-parts":[[2026,4,9]],"date-time":"2026-04-09T00:00:00Z","timestamp":1775692800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Yunnan Sci-Tech Talent and Platform Plan Project","award":["202605AF350036"],"award-info":[{"award-number":["202605AF350036"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IJGI"],"abstract":"<jats:p>Worldwide visual geo-localization aims to predict the geographic coordinates of an image capture location from visual content alone, posing unique challenges due to the vast scale of the Earth\u2019s surface and pervasive visual ambiguity across distant regions. Existing approaches face distinct limitations as follows: retrieval-based methods demand massive geo-tagged databases and scale poorly; alignment-based models lack interpretability and are vulnerable to visually similar scenes; and large vision-language models (LVLMs) offer semantic reasoning but suffer from hallucination. A natural solution is retrieval-augmented generation (RAG), yet we observe that directly injecting retrieved candidates as context causes severe context poisoning. To address this, we propose HybridGeo, a dual-stream late-fusion framework that decouples retrieval from reasoning. A retrieval stream applies continuous alignment with spatial\u2013semantic clustering to produce stable regional anchors; a reasoning stream performs context-free Chain-of-Thought inference to yield an independent coordinate estimate. The two streams are fused only at the decision stage via a spatial\u2013consistency module that triggers weighted averaging under agreement or confidence-based arbitration under conflict. Experiments on Im2GPS3k show that HybridGeo achieves 73.89% Country@750km accuracy, outperforming the retrieval baseline by 7.27% and 8.23%, and surpassing both VLM-only and RAG baselines. These results demonstrate that late fusion effectively avoids context poisoning while enabling complementary benefits from both streams.<\/jats:p>","DOI":"10.3390\/ijgi15040163","type":"journal-article","created":{"date-parts":[[2026,4,10]],"date-time":"2026-04-10T07:55:12Z","timestamp":1775807712000},"page":"163","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Synergizing Retrieval and CoT Reasoning via Spatial Consensus for Worldwide Visual Geo-Localization"],"prefix":"10.3390","volume":"15","author":[{"given":"Yong","family":"Tang","sequence":"first","affiliation":[{"name":"Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100094, China"},{"name":"University of Chinese Academy of Sciences, Beijing 100049, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Jianhua","family":"Gong","sequence":"additional","affiliation":[{"name":"Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100094, China"},{"name":"University of Chinese Academy of Sciences, Beijing 100049, China"},{"name":"Zhejiang-CAS Application Center for Geoinformatics, Jiaxing 314100, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2642-2194","authenticated-orcid":false,"given":"Yi","family":"Li","sequence":"additional","affiliation":[{"name":"Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100094, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Jieping","family":"Zhou","sequence":"additional","affiliation":[{"name":"Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100094, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Jun","family":"Sun","sequence":"additional","affiliation":[{"name":"Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100094, China"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1968","published-online":{"date-parts":[[2026,4,9]]},"reference":[{"key":"ref_1","first-page":"5","article-title":"Primary exploration of geographic metaverse from the perspective of virtual geographic environment","volume":"28","author":"Gong","year":"2024","journal-title":"Natl. Remote Sens. Bull."},{"key":"ref_2","doi-asserted-by":"crossref","unstructured":"Kendall, A., Grimes, M., and Cipolla, R. (2015, January 7\u201313). PoseNet: A convolutional network for real-time 6-DOF camera relocalization. Proceedings of the IEEE International Conference on Computer Vision, Santiago, Chile.","DOI":"10.1109\/ICCV.2015.336"},{"key":"ref_3","doi-asserted-by":"crossref","unstructured":"Huh, M., Liu, A., Owens, A., and Efros, A.A. (2018, January 8\u201314). Fighting fake news: Image splice detection via learned self-consistency. Proceedings of the European Conference on Computer Vision (ECCV), Munich, Germany.","DOI":"10.1007\/978-3-030-01252-6_7"},{"key":"ref_4","doi-asserted-by":"crossref","unstructured":"Middelberg, S., Sattler, T., Untzelmann, O., and Kobbelt, L. (2014, January 6\u201312). Scalable 6-DOF localization on mobile devices. Proceedings of the European Conference on Computer Vision, Zurich, Switzerland.","DOI":"10.1007\/978-3-319-10605-2_18"},{"key":"ref_5","first-page":"53878","article-title":"Automatic geolocation estimation of images","volume":"6","author":"Cao","year":"2018","journal-title":"IEEE Access"},{"key":"ref_6","doi-asserted-by":"crossref","unstructured":"Hays, J., and Efros, A.A. (2008, January 23\u201328). IM2GPS: Estimating geographic information from an image collection. Proceedings of the 2008 IEEE Conference on Computer Vision and Pattern Recognition, Anchorage, AK, USA.","DOI":"10.1109\/CVPR.2008.4587784"},{"key":"ref_7","doi-asserted-by":"crossref","unstructured":"Arandjelovi\u0107, R., Gronat, P., Torii, A., Pajdla, T., and Sivic, J. (2016, January 27\u201330). NetVLAD: CNN architecture for weakly supervised place recognition. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA.","DOI":"10.1109\/CVPR.2016.572"},{"key":"ref_8","doi-asserted-by":"crossref","unstructured":"Weyand, T., Kostrikov, I., and Philbin, J. (2016, January 11\u201314). Planet-photo geolocation with convolutional neural networks. Proceedings of the European Conference on Computer Vision, Amsterdam, The Netherlands.","DOI":"10.1007\/978-3-319-46484-8_3"},{"key":"ref_9","unstructured":"Luo, C., Li, L., Wen, K., Yan, X., Zhang, A., and Nie, L. (2023, January 10\u201316). GeoCLIP: Clip-Inspired Alignment between Locations and Images for Effective Worldwide Geo-localization. Proceedings of the Advances in Neural Information Processing Systems, New Orleans, LA, USA."},{"key":"ref_10","doi-asserted-by":"crossref","unstructured":"Haas, L., Schurholt, K., Klemmer, K., and Borth, D. (2024, January 17\u201321). PIGEON: Predicting Image Geolocations. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition, Seattle, WA, USA.","DOI":"10.1109\/CVPR52733.2024.01225"},{"key":"ref_11","unstructured":"OpenAI (2026, March 30). GPT-4V(ision) System Card. Available online: https:\/\/openai.com\/research\/gpt-4v-system-card."},{"key":"ref_12","unstructured":"DeepMind, G. (2026, March 30). Gemini: A Family of Highly Capable Multimodal Models. Available online: https:\/\/deepmind.google\/technologies\/gemini\/."},{"key":"ref_13","unstructured":"Li, L., Ye, Y., Jiang, B., and Zeng, W. (2024). GeoReasoner: Geo-localization with Reasoning in Street Views using a Large Vision-Language Model. arXiv."},{"key":"ref_14","unstructured":"Zhang, Z., Wang, R.E., Kabir, T., and Boyd-Graber, J. (2025). NAVIG: Natural Language-guided Analysis with Vision Language Models for Image Geo-localization. arXiv."},{"key":"ref_15","unstructured":"Wang, Y., Liu, Z., Wang, Z., Liu, P., Hu, H., and Artzi, Y. (2025). GeoVista: Web-Augmented Agentic Visual Reasoning for Geolocalization. arXiv."},{"key":"ref_16","doi-asserted-by":"crossref","unstructured":"Berton, G., Masone, C., and Caputo, B. (2022, January 19\u201324). Rethinking visual geo-localization for large-scale applications. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition, New Orleans, LA, USA.","DOI":"10.1109\/CVPR52688.2022.00483"},{"key":"ref_17","doi-asserted-by":"crossref","unstructured":"Ali-bey, A., Chaib-draa, B., and Gigu\u00e8re, P. (2023, January 3\u20137). MixVPR: Feature mixing for visual place recognition. Proceedings of the IEEE\/CVF Winter Conference on Applications of Computer Vision, Waikoloa, HI, USA.","DOI":"10.1109\/WACV56688.2023.00301"},{"key":"ref_18","unstructured":"Zhu, S., Yang, T., and Chen, C. (2023, January 18\u201322). TransLocator: Global-Scale Visual Geo-Localization with Transformers. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition, Vancouver, BC, Canada."},{"key":"ref_19","unstructured":"Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Goyal, N., K\u00fcttler, H., Lewis, M., Yih, W.t., and Rockt\u00e4schel, T. (2020, January 6\u201312). Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks. Proceedings of the Advances in Neural Information Processing Systems, Virtual Conference."},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"205","DOI":"10.21105\/joss.00205","article-title":"hdbscan: Hierarchical density based spatial clustering of applications with noise","volume":"2","author":"McInnes","year":"2017","journal-title":"J. Open Source Softw."},{"key":"ref_21","unstructured":"(2026, March 30). StreetView 5M Dataset. Available online: https:\/\/github.com\/gastruc\/osv5m."},{"key":"ref_22","doi-asserted-by":"crossref","unstructured":"Muller-Budack, E., Pustu-Iren, K., and Ewerth, R. Geolocation Estimation of Photos Using a Hierarchical Model and Scene Classification. Computer Vision\u2014ECCV 2018, Proceedings of the 15th European Conference on Computer Vision (ECCV), Munich, Germany, 8\u201314 September 2018, Springer.","DOI":"10.1007\/978-3-030-01258-8_35"},{"key":"ref_23","doi-asserted-by":"crossref","unstructured":"Seo, P., Weyand, T., Sim, J., and Han, B. CPlaNet: Enhancing Image Geo-localization by Combinatorial Partitioning of Maps. Computer Vision\u2014ECCV 2018, Proceedings of the 15th European Conference on Computer Vision (ECCV), Munich, Germany, 8\u201314 September 2018, Springer.","DOI":"10.1007\/978-3-030-01249-6_33"}],"container-title":["ISPRS International Journal of Geo-Information"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2220-9964\/15\/4\/163\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,4,17]],"date-time":"2026-04-17T08:22:57Z","timestamp":1776414177000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2220-9964\/15\/4\/163"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,4,9]]},"references-count":23,"journal-issue":{"issue":"4","published-online":{"date-parts":[[2026,4]]}},"alternative-id":["ijgi15040163"],"URL":"https:\/\/doi.org\/10.3390\/ijgi15040163","relation":{},"ISSN":["2220-9964"],"issn-type":[{"value":"2220-9964","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,4,9]]}}}