In the MiTrAs project new methods for monitoring, early detection and trend analysis are being developed. The fusion of information from satellite images, open source and social media from outside Europe is used in the research project to assess developments in dedicated application areas. The research project MiTrAs focuses on the following main research topics:
• Multi-temporal satellite image analysis for object recognition of vehicles and groups of people by developing automated deep-learning analysis methods (e.g. convolutional neural networks), applied to panchromatic and multispectral image data with high resolution. New technologies such as high-resolution radar satellites and the use of satellite video are being investigated and evaluated for the first time for their suitability for the detection of small objects.
• New deep learning methods and algorithms are also being developed for the analysis and processing of heterogeneous data from open source sources and social media communication. An essential factor here is to identify the events relevant to the above-mentioned areas of application and to check them with regard to their credibility, topicality and geographical origin.
• Algorithms and interpretation concepts for the fusion of the spatially and temporally different sources of information in order to be able to correlate the analysed observations according to plausibility and relevance.The research results are presented and evaluated by means of a simple visualization.
Project Duration: January 2019 – August 2020
Funding: FFG, KIRAS
Project Coordination: AIT Austrian Institute of Technology GmbH
Project Partners: Universität für Weiterbildung Krems (Donau-Universität Krems), Zentralanstalt für Meteorologie und Geodynamik, SYNYO GmbH, Bundesministerium für Inneres BM.I, Research Institute AG & Co KG, Bundesministerium für Landesverteidigung
Contact: Barbara Salcher, Maegan Hendow