BEGIN:VCALENDAR VERSION:2.0 PRODID:-//pretalx//talks.osgeo.org//foss4g-2026//talk//FTA7PT BEGIN:VTIMEZONE TZID:JST BEGIN:STANDARD DTSTART:20000101T000000 RRULE:FREQ=YEARLY;BYMONTH=1 TZNAME:JST TZOFFSETFROM:+0900 TZOFFSETTO:+0900 END:STANDARD END:VTIMEZONE BEGIN:VEVENT UID:pretalx-foss4g-2026-FTA7PT@talks.osgeo.org DTSTART;TZID=JST:20260901T133000 DTEND;TZID=JST:20260901T140000 DESCRIPTION:1 Introduction and Objectives\nThe large-scale development and maintenance of high-quality 3D city models remain a key challenge in the d igital transformation of urban planning and management. In Japan\, Project PLATEAU\, launched by the Ministry of Land\, Infrastructure\, Transport a nd Tourism (MLIT) in 2020\, promotes the creation\, utilization\, and open dissemination of 3D city models as foundational data for smart city initi atives. Despite steady progress\, the time and cost required to generate d etailed models—particularly at Level of Detail (LOD) 2 or higher—conti nue to pose significant constraints\, as such models are still largely pro duced through labor-intensive manual processes.\nProject PLATEAU aims to c omplete 3D city models for 500 municipalities by fiscal year 2027\, with r oughly two-thirds achieved by fiscal year 2025. Achieving full coverage wi ll require substantial reductions in production cost. In addition\, 3D cit y models quickly become outdated due to ongoing urban changes such as cons truction and demolition\, making periodic updates essential. However\, hig h update costs risk limiting their timely maintenance\, underscoring the n eed for scalable and automated solutions that can reduce costs while meeti ng strict quality requirements for municipal use.\n\nIn response\, this st udy presents the development and validation of a beta version of an AI-dri ven automated modeling tool\, named AI City Model Maker\, designed to supp ort the generation and updating of 3D city models at LOD2 and above. The p rimary objective of the tool is not to fully replace human operators\, but to significantly reduce manual workload by automating substantial portion s of the modeling process\, thereby enabling cost-effective and repeatable production workflows.\n\n2 Requirements \nThe tool is designed to comply with the “Standard Data Product Specification for 3D City Models” esta blished by PLATEAU\, which is based on the CityGML standard and aligned wi th ISO 19100 series quality requirements. This specification defines strin gent criteria for positional accuracy\, completeness\, logical consistency \, and thematic accuracy\, reflecting the expectation that 3D city models will be used in official municipal operations such as urban planning\, dis aster management\, and infrastructure assessment. For example\, when model s are generated from Level 2500 source data\, horizontal positional accura cy must be within a standard deviation of 1.75 meters\, and vertical accur acy within 0.66 meters\, while logical and thematic errors are not permitt ed. Meeting these requirements through fully automated processes remains c hallenging with current technology.\n\n3 Architecture and Applied Technolo gies\nThe proposed tool addresses this challenge through a hybrid workflow that combines AI-based automation with targeted human editing. It takes a s input quality-assured datasets commonly available to Japanese local gove rnments\, including aerial imagery\, digital surface models\, building foo tprints\, and digital elevation models. Using these inputs\, the tool auto matically generates building models at LOD1 and LOD2\, road models at LOD1 and LOD2\, and city furniture and vegetation models at LOD3. Outputs are provided in standard formats such as CityGML\, enabling seamless integrati on into existing geospatial workflows and open data ecosystems.\n\nFrom a technical perspective\, the tool integrates multiple open-source and state -of-the-art AI technologies. Building model generation is based on an OSS “Auto-Create-bldg-lod2-tool”\, enhanced with deep learning models such as convolutional and transformer-based architectures. Road extraction and modeling leverage semantic segmentation techniques\, while city furniture and vegetation modeling combine point cloud analysis\, object detection\, clustering\, and parametric modeling. The system is implemented in both c loud-based and desktop-based configurations\, allowing flexibility in depl oyment depending on data volume\, network conditions\, and organizational constraints.\n\n4 Implementation and Validation\nTo ensure practical relev ance\, the beta version of the tool has been deployed for pilot testing at six surveying and construction consulting companies in Japan. These compa nies are actively incorporating the tool into their existing 3D city model production pipelines\, and regular feedback is being collected to guide f urther development. This real-world testing distinguishes the present work from many prior studies\, which often remain at the proof-of-concept or r esearch prototype stage.\nClear performance metrics and automation targets were defined during implementation. These include overall cost reduction across an entire city model\, cost reduction at the individual feature lev el\, and automation rates required to achieve these reductions. The long-t erm target is an overall cost reduction of 30–50%\, assuming that some d egree of manual editing remains necessary to meet quality standards. Featu re-level targets vary by object type\, reflecting differences in geometric complexity and data availability.\n\nEvaluation results at the beta stage indicate mixed performance. For building models at LOD2\, the automation rate for high-quality outputs remains well below the target\, highlighting the difficulty of reliably reconstructing complex roof geometries from st andard aerial imagery and elevation data. In contrast\, road models and ci ty furniture and vegetation models generally met or approached their respe ctive targets\, although performance varied depending on local conditions such as urban density and road structure. These results suggest that while AI-based automation is already effective for certain feature types\, furt her improvements are required for complex building geometries.\n\n5 Conclu sion\nDespite not yet achieving all targets\, the beta version demonstrate s the feasibility of a tool-oriented approach to AI-driven 3D city model g eneration. By focusing on usability\, integration into existing workflows\ , and iterative improvement based on practitioner feedback\, the proposed tool represents a step toward the social implementation of automated 3D ci ty modeling technologies. In particular\, it supports the expansion of ope n 3D city model coverage by lowering the barrier to production and updates \, which aligns closely with the goals of both Project PLATEAU and the bro ader open geospatial community.\n\nFuture work will focus on improving mod eling accuracy for existing feature types and extending automation to more detailed representations\, including building and road models at LOD3. Th rough continued collaboration with industry users and further advances in AI and open-source geospatial technologies\, the proposed approach aims to contribute to sustainable\, scalable\, and openly accessible 3D city mode l ecosystems. DTSTAMP:20260717T234903Z LOCATION:Cosmos1 SUMMARY:Scaling Open 3D City Models: Implementation and Validation of an AI -driven Automated Generation Tool “AI City Model Maker beta version” - Mayumi Mizobuchi URL:https://talks.osgeo.org/foss4g-2026/talk/FTA7PT/ END:VEVENT END:VCALENDAR