Prof. Demir demonstrates AI applications in earth observation research
How do you analyze a treasure trove of data that grows by 12 terabytes daily? How do you use AI to leverage satellite imagery for environmental protection? Prof. Dr. Begüm Demir heads the BIFOLD research group Big Data Analytics for Earth Observation. In the Tagesspiegel TU Berlin supplement , she presents her research: among other things, an analysis and information system for the evaluation of satellite data from "Sentinel-1" and "Sentinel-2", as well as the reference image database "BigEarthNet" and the search engine "EarthQube." Furthermore, Begüm Demir explains the advantages of Explainable Artificial Intelligence (XAI) in Earth Observation research. This is illustrated by the project "TreeSatAI," a cooperation with the "Geoinformation in Environmental Planning" department at TU Berlin, led by Prof. Dr. Birgit Kleinschmit, LiveEO GmbH, LUP – Luftbild Umwelt Planung GmbH, Deutsches Zentrum für Künstliche Intelligenz GmbH and Vision Impulse GmbH.
Supplement of the Technische Universität Berlin in cooperation with the Tagesspiegel (in German), October 18th, 2023, page 3: "Forschung von ganz oben"
Publication: Steve Ahlswede, Christian Schulz, Christiano Gava, Patrick Helber, Benjamin Bischke, Michael Förster, Florencia Arias, Jörn Hees, Begüm Demir, and Birgit Kleinschmit: TreeSatAI Benchmark Archive: a multi-sensor, multi-label dataset for tree species classification in remote sensing
DOI: https://doi.org/10.5194/essd-15-681-2023
The TreeSatAI Benchmark Archive is based on reference data from forest administration data in Germany, the dataset aims to gather multi-sensor and multi-label information for the classification of 20 tree species in central Europe. The TreeSatAI Benchmark Archive consists of 50 381 image patches from aerial, Sentinel-2 (S2), and Sentinel-1 (S1) imagery, with a range of 212 to 6591 individual samples per class.