Meet us at ACM SIGSPATIAL 2025

The ACM SIGSPATIAL conference is one of the most respected venues for geospatial data science. This year, we are part of it again and looking forward to the exchange.

We contribute to the mission of this conference by

• organizing the GeoSearch workshop still awaiting your submissions,
• co-chairing the demo track
• presenting our latest research…

This time, we are present in the main conference with the following papers:

The paper Enhancing Contrastive Learning for Geolocalization by Discovering Hard Negatives on Semivariograms introduces a novel spatially regularized contrastive learning strategy that integrates a semivariogram in order to relate visual similarity and spatial distance with each other. In this way, more meaningful hard negative samples can be made available during training. The strategy has been integrated with GeoCLIP and evaluated on the OSV5M dataset, demonstrating that explicitly modeling spatial priors improves image-based geo-localization performance, particularly at finer granularity.

The paper SHM-DB - A Shared Memory Database for FAIR Geospatial Algorithm Development introduces the shared memory database SHM-DB and demonstrates how this can be used for developing point cloud processing algorithms on comparably large datasets with improved software quality. Beyond allowing loose coupling and flexible choice of tooling, this system provides real-time visualization of the inner workings of algortihms with the potential to save a lot of researchers time as erroneous runs (e.g., with bad parameters) can be identified early.

We look forward to meeting you in Minnesota!