As the geospatial industry evolves, new technologies are rapidly reshaping how data is captured, processed, and visualized. This session highlights a range of forward-looking innovations that push the boundaries of what’s possible in mapping and spatial analysis. Learn how open-source AI tools like Meta’s Segment Anything Model (SAM) are being applied to improve object detection and accuracy in drone imagery, making advanced analysis more accessible than ever. Discover how breakthroughs in server-side rendering are solving one of the biggest challenges in geospatial data—visualizing and sharing massive 3D models efficiently across the web. And explore how sensor-equipped vehicles, such as those in the Tesla fleet, could soon transform everyday transportation into a powerful mapping network capable of producing survey-grade results. Together, these presentations illustrate a future where the lines between AI, visualization, and mobility continue to blur, unlocking faster, smarter, and more scalable geospatial workflows.
The following presentations will be shared in this session:
Performing Object Detection with Meta’s Segment Anything Model (SAM)
Presented by Nicholas MCarty, Upskilled Consulting
High-resolution orthomosaics created from nadir drone imagery are a cornerstone of modern geospatial analysis, enabling precise mapping of real-world features across large areas. This session explores how mask segmentation using Meta’s open-source Segment Anything Model (SAM) can be applied to orthophotos to identify objects of interest, calculate polygon centroids, and compare results against ground control points to quantify spatial accuracy. Attendees will learn how the Python package orthomasker streamlines this workflow, producing geo-located polygons in GeoJSON format with approximately 20% greater spatial accuracy and 400% higher intersection-over-union (IoU) than results from costly proprietary software. The session will highlight practical applications, demonstrate reproducible workflows, and showcase how open-source tools can deliver superior results for mapping, monitoring, and analysis projects.
Solving Bottlenecks of Large-Scale 3D Data Visualization with Server Side Rendering Technology
Presented by Joon Heo, Conworth, Inc.
Large-scale 3D data is increasingly being utilized across various fields, enabled by more precise and efficient acquisition technologies and the integration of rich attributes, including dynamic information. However, its immense size and complexity create significant challenges for internet-based visualization and act as a bottleneck in data delivery to end users, ultimately limiting its full potential. This presentation introduces a visualization solution based on Server Side Rendering (SSR) that addresses this critical bottleneck. It also highlights practical applications of the Conworth Engine, which implements this technology in the fields of AEC, EPC, asset/facility management, digital twin of software-defined factory (SDF), and digital cultural heritage experiences.
Utilizing Tesla Sensor Suite for High-Fidelity Mobile Mapping
Presented by Rami Tamimi, The Survey School
With millions of sensor-equipped vehicles on the road, the Tesla fleet represents an unprecedented opportunity for geospatial data collection. This presentation explores the potential and the practical challenges of leveraging Tesla’s integrated cameras for professional mapping applications. We will present a case study that bypasses current data access limitations by pairing a vehicle with an external, high-precision GNSS receiver. This workflow demonstrates how to generate dimensionally accurate, 3D-ready imagery for infrastructure surveys, asset management, and digital twin creation, turning a consumer vehicle into a scalable mapping tool.
Spatial AI in Your Pocket: Redefining Mobile 3D Capture for Real-World Field and Robotics Workflows
Presented by Tianyue Yu, Q3D Sensing
High-quality 3D capture has traditionally relied on bulky and expensive equipment operated by experts, limiting its utility in dynamic, real-world environments such as infrastructure inspections, construction sites, industrial facilities and autonomous system development. This session introduces a new category of 3D reality capture: compact, AI-enabled, palm-sized devices, designed for real time capture in the field, and many other use cases wherever rich, accurate depth data is needed. We’ll share insights from deployments across construction, robotics, and design workflows, showcasing how these tools unlock new possibilities for in-field spatial data collection, without the constraints of legacy systems. We’ll explore how advances in miniaturized sensing, spatial AI, and edge computing are shifting 3D capture from a specialized, expensive process to an accessible, affordable, real-time workflow that removes scanning complexity so that anyone onsite is capable of mapping existing conditions.
