Today, digital geodata is essential in many scientific disciplines because it clearly shows objects with a spatial reference, which play a central role in our digital world. Due to the abundance of geodata with different representations, the demand to combine the data with each other is also growing. To make this possible, researchers led by Prof. Boris Jutzi and Dr. Olaf Wysocki created a holistic set of currently 32 geodata with a detailed digital representation of the TUM university campus in Munich's city center.
Doctoral candidates in geodesy, computer science and mathematics launched the initiative in 2021. They measure, reference, transform and visualize data and combine the parts into a new whole - with the georeferencing of the data being the link.
Olaf Wysocki, currently a postdoctoral researcher, is coordinating the interdisciplinary joint project. "We chose the TUM campus as a geodetic reference system because it enables a complex urban scenario for various method tests, for example with buildings of different architectural styles. On this basis, methods and applications can be compared and evaluated, such as pose estimation, vehicle positioning, building reconstruction, semantic segmentation, facade inpainting, thermal point cloud projection, driving & biking simulator, solar potential analysis and novel view synthesis. There is a wide range of possible applications in science and beside this to keep the data up to date. That's why our colleague Benedikt Schwab has developed a website where the data is updated regularly", says Wysocki.
The results of fruitful interdisciplinary cooperation:
- 32 multitemporal, multimodal, high-resolution data sets with up to millimeter-precise representation of interior and exterior scenes of the TUM campus.
- large-scale coverage of 100,000 square meters of the TUM campus and its surroundings.
- diverse data sources: In addition to stationary sensors, mobile sensor platforms such as vehicles, drones, aircraft and satellites are also used. These images are used to create high-quality high-level derivatives in the form of orthoimages, point clouds, semantic 3D city models, high-resolution maps and a wide range of geoinformation.
- diverse research approaches and applications: The unique setup of TUM2TWIN is already being used in a variety of scientific projects, from coregistration to building reconstruction to implicit scene representation.
"What makes our holistic approach so unique is the temporal and spatial unification of all data sets into a common coordinate system to form a digital twin. Whether point clouds, satellite images, UAS laser scans or 3D models: the freely available TUM2TWIN data in a multitemporal and multimodal representation is unique," emphasizes Prof. Boris Jutzi.
Links:
Open Source Project TUM2twin
Prof. Boris Jutzi
Dr. Olaf Wysocki
Professorship of Photogrammetry and Remote Sensing
Pre-print: TUM2TWIN: Introducing the Large-Scale Multimodal Urban Digital Twin Benchmark Dataset
Cooperations:
Professorship of Photogrammetry and Remote Sensing
Chair of Geoinformatics
Chair of Traffic Engineering and Control
Chair of Data Science in Earth Observation
Chair of Engineering Geodesy
3D AI Laboratory
Visual Computing & Artificial Intelligence Lab
Chair of Computing in Civil and Building Engineering
Professorship of Green Technologies in Landscape Architecture
Professorship of Remote Sensing Applications
Chair of Architectural Informatics
Department of Geoinformatics (University of Applied Sciences Munich)
Remote Sensing Technology Institute, Photogrammetry and Image Analysis Department, German Aerospace Center (DLR)