We invite contributions on Geospatial Artificial Intelligence (GeoAI) for Earth Observation, from foundation models and spatial reasoning on the ground to AI running on board satellites and UAVs. The workshop welcomes methodological and application-oriented works.
Advances in AI and cloud computing are changing how Earth Observation (EO) data are processed and interpreted. Deep learning, multimodal learning, geospatial foundation models, and large language models let EO systems move beyond pixel-level classification towards generalizable, transferable, and knowledge-aware geospatial intelligence. Combining data-driven learning with physical knowledge, as in physics-informed AI, further improves the consistency and interpretability of results.At the same time, satellite constellations collect terabytes of optical, SAR, and hyperspectral data every day. Sending raw data to the ground creates bandwidth bottlenecks and latency, which matter most in disaster response. Lightweight neural accelerators, model compression and graph-based learning now make it possible to analyze data directly on board and send compact, actionable information to the ground.These developments raise open scientific questions: spatial and temporal generalization, uncertainty, explainability, data heterogeneity, computational efficiency, physical consistency, reproducibility, and the responsible use of AI. The workshop brings together remote sensing scientists, geospatial experts, computer scientists, and domain specialists to discuss new methods, applications, and future directions. Contributions addressing societal challenges, such as climate change, biodiversity, sustainable urban development, natural hazards, food security, and ecosystem resilience, are particularly welcome.
List of topics
Foundation models and learning methods
- Geospatial foundation models: pre-training, adaptation and evaluation for EO
- Spatial machine learning and deep learning; self-supervised and transfer learning
- Large language models and vision–language models for geospatial applications
- Multimodal and multi-sensor data fusion (optical, SAR, LiDAR, hyperspectral, thermal)
- Physics-informed and physics-based AI: physics-informed neural networks (PINNs), hybrid physics–ML models and AI-based inversion of physical models.
Spatial intelligence and digital twins
- Spatial reasoning and spatio-temporal modelling, including graph neural networks
- Digital twins and 3D spatial intelligence
- Explainable, trustworthy and uncertainty-aware GeoAI
- Benchmarks, datasets and reproducibility
Edge AI and on-board processing
- Lightweight models and hardware acceleration: quantization, pruning and compilation for satellite and UAV processors
- Semantic compression and low-bandwidth downlink of EO information
- In-orbit demonstrations and lessons learned (e.g. Φ-Sat, OPS-SAT, NewSpace missions)
Applications
- Real-time disaster response and emergency mapping (floods, wildfires, earthquakes)
- Land use and land cover, urban systems, agriculture, forestry and environmental monitoring
- GeoAI for the UN Sustainable Development Goals and socio-economic mapping
Organizers:
Dr. Mohammad Kakooei, Geomatics, Karlstad University, Sweden, mohammad.kakooei@kau.se
Dr. Jan Haas, Geomatics, Karlstad University, Sweden, jan.haas@kau.se
Maamar Omar Arbouz, EdgeOrbit AI, arbouzmaamar@gmail.com
Dr. Roland Perko, JOANNEUM RESEARCH, Austria, roland.perko@joanneum.at
Jonas Viehweger, JOANNEUM RESEARCH, Austria, jonas.viehweger@joanneum.at
Prof. Fabio Del Frate, University of Rome “Tor Vergata”, fabio.del.frate@uniroma2.it
Scientific Committee:
- Prof. Luka Čehovin Zajc, University of Ljubljana, Slovenia, luka.cehovin@fri.uni-lj.si
- Dr. Christian Ayala Lauroba, Tracasa Instrumental, Spain, c.ayala@itracasa.es
- Dr. Sead Mustafic, JOANNEUM RESEARCH, Austria, Sead.Mustafic@joanneum.at
- Prof. Bernd Resch, University of Salzburg, Austria, bernd.resch@plus.ac.at
- Prof. Jan Dirk Wegner, University of Zurich, Switzerland, jan.wegner@geod.baug.ethz.ch
- Dr. Can Atalay, Bulent Ecevit University, Türkiye, canatalay@beun.edu.tr
- Dr. Ksenia Bittner, German Aerospace Center, Germany, ksenia.bittner@dlr.de
- Dr. Meisam Amani, Remote Digital twin Inc. (RDT), Canada, M.Amani@remotedigitaltwin.com
- Dr. Andrea Nascetti, KTH Royal Institute of Technology, Sweden, nascetti@kth.se
- Dr. Olof Mogren, Lund University and RISE Research Institutes, Sweden, olof.mogren@ri.se
- Dr. Davide De Santis, University of Rome “Tor Vergata”, davide.de.santis@uniroma2.it