{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/bev-cv-birds-eye-view-transform-for-cross","title":"BEV-CV: Birds-Eye-View Transform for Cross-View Geo-Localisation","arxiv_id":"2312.15363","date":"2023-12-23","proceeding":null,"authors":["Tavis Shore","Simon Hadfield","Oscar Mendez"],"abstract":"Cross-view image matching for geo-localisation is a challenging problem due to the significant visual difference between aerial and ground-level viewpoints. The method provides localisation capabilities from geo-referenced images, eliminating the need for external devices or costly equipment. This enhances the capacity of agents to autonomously determine their position, navigate, and operate effectively in GNSS-denied environments. Current research employs a variety of techniques to reduce the domain gap such as applying polar transforms to aerial images or synthesising between perspectives. However, these approaches generally rely on having a 360{\\deg} field of view, limiting real-world feasibility. We propose BEV-CV, an approach introducing two key novelties with a focus on improving the real-world viability of cross-view geo-localisation. Firstly bringing ground-level images into a semantic Birds-Eye-View before matching embeddings, allowing for direct comparison with aerial image representations. Secondly, we adapt datasets into application realistic format - limited Field-of-View images aligned to vehicle direction. BEV-CV achieves state-of-the-art recall accuracies, improving Top-1 rates of 70{\\deg} crops of CVUSA and CVACT by 23% and 24% respectively. Also decreasing computational requirements by reducing floating point operations to below previous works, and decreasing embedding dimensionality by 33% - together allowing for faster localisation capabilities.","url_abs":"https://arxiv.org/abs/2312.15363v2","url_pdf":"https://arxiv.org/pdf/2312.15363v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"bev-cv-birds-eye-view-transform-for-cross","repo_url":"https://github.com/tavisshore/BEV-CV","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"camera-localization","task_name":"Camera Localization"},{"task_slug":"cross-view-geo-localisation","task_name":"Cross-View Geo-Localisation"},{"task_slug":"image-retrieval","task_name":"Image Retrieval"},{"task_slug":"image-based-localization","task_name":"Image-Based Localization"},{"task_slug":"navigate","task_name":"Navigate"},{"task_slug":"outdoor-localization","task_name":"Outdoor Localization"},{"task_slug":"retrieval","task_name":"Retrieval"},{"task_slug":"visual-localization","task_name":"Visual Localization"},{"task_slug":"geo-localization","task_name":"geo-localization"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/cross-view-geo-localisation-on-cvusa-70","task":"Cross-View Geo-Localisation","dataset":"CVUSA 70","model":"BEV-CV","rank_in_archive_order":1,"of":1,"metrics":{"Top-1":"27.4","Top-1%":"90.94","Top-10":"64.47","Top-5":"52.94"},"uses_additional_data":false},{"leaderboard":"/sota/cross-view-geo-localisation-on-cvusa-90","task":"Cross-View Geo-Localisation","dataset":"CVUSA 90","model":"DSM","rank_in_archive_order":1,"of":12,"metrics":{"Top-1":"33.66"},"uses_additional_data":false},{"leaderboard":"/sota/cross-view-geo-localisation-on-cvusa-90","task":"Cross-View Geo-Localisation","dataset":"CVUSA 90","model":"BEV-CV","rank_in_archive_order":2,"of":12,"metrics":{"Top-1":"32.11","Top-1%":"92.99","Top-10":"69.06","Top-5":"58.36"},"uses_additional_data":false},{"leaderboard":"/sota/cross-view-geo-localisation-on-cvusa-90","task":"Cross-View Geo-Localisation","dataset":"CVUSA 90","model":"L2LTR","rank_in_archive_order":3,"of":12,"metrics":{"Top-1":"25.21"},"uses_additional_data":false},{"leaderboard":"/sota/cross-view-geo-localisation-on-cvusa-90","task":"Cross-View Geo-Localisation","dataset":"CVUSA 90","model":"GAL","rank_in_archive_order":4,"of":12,"metrics":{"Top-1":"22.54"},"uses_additional_data":false},{"leaderboard":"/sota/cross-view-geo-localisation-on-cvusa-90","task":"Cross-View Geo-Localisation","dataset":"CVUSA 90","model":"TransGeo [Zhu2022TransGeoTI]","rank_in_archive_order":5,"of":12,"metrics":{"Top-1":"21.96"},"uses_additional_data":false},{"leaderboard":"/sota/cross-view-geo-localisation-on-cvusa-90","task":"Cross-View Geo-Localisation","dataset":"CVUSA 90","model":"GeoDTR","rank_in_archive_order":6,"of":12,"metrics":{"Top-1":"15.21","Top-10":"52.27","Top-5":"39.32"},"uses_additional_data":false},{"leaderboard":"/sota/cross-view-geo-localisation-on-cvusa-90","task":"Cross-View Geo-Localisation","dataset":"CVUSA 90","model":"CVFT","rank_in_archive_order":7,"of":12,"metrics":{"Top-1":"4.8"},"uses_additional_data":false},{"leaderboard":"/sota/cross-view-geo-localisation-on-cvusa-90","task":"Cross-View Geo-Localisation","dataset":"CVUSA 90","model":"CVM","rank_in_archive_order":8,"of":12,"metrics":{"Top-1":"2.76"},"uses_additional_data":false},{"leaderboard":"/sota/cross-view-geo-localisation-on-cvusa-90","task":"Cross-View Geo-Localisation","dataset":"CVUSA 90","model":"GeoDTR [zhang2023crossview]","rank_in_archive_order":9,"of":12,"metrics":{"Top-1%":"88.72"},"uses_additional_data":false},{"leaderboard":"/sota/cross-view-geo-localisation-on-cvusa-90","task":"Cross-View Geo-Localisation","dataset":"CVUSA 90","model":"TransGeo","rank_in_archive_order":10,"of":12,"metrics":{"Top-1%":"86.8","Top-10":"56.49","Top-5":"45.35"},"uses_additional_data":false},{"leaderboard":"/sota/cross-view-geo-localisation-on-cvusa-90","task":"Cross-View Geo-Localisation","dataset":"CVUSA 90","model":"L2LTR [Yang2021CrossviewGW]","rank_in_archive_order":11,"of":12,"metrics":{"R@5":"51.9"},"uses_additional_data":false},{"leaderboard":"/sota/cross-view-geo-localisation-on-cvusa-90","task":"Cross-View Geo-Localisation","dataset":"CVUSA 90","model":"DSM [Shi2020WhereAI]","rank_in_archive_order":12,"of":12,"metrics":{"R@5":"51.7"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}