Severity level assessment from semantically fused video content analysis for physical threat detection in ground segments of space systems
Gerasimos S. Antzoulatos, Orfanidis Georgios, Panagiotis Giannakeris, Tzanetis Giorgos, Kampilis-Stathopoulos Grigorios, Nikolaos Kopalidis, Ilias Gialampoukidis, Stefanos Vrochidis, Kompatsiaris Ioannis · Zenodo (CERN European Organization for Nuclear Research) · 2021
Disaster risks related to natural hazards are evolving gradually, albeit accelerating over time, the human-made and cyber threats are changing rapidly exploiting the increasing progress in technologies and the complex, highly interlinked, modern environment of critical infrastructures. Therefore, as these threats have been intensifying, the actions to strengthen the resilience of critical infrastructures should be step up, by understanding their complex systems as well as the multi-risks nature. In this landscape, the aim of this work focuses on proposing a framework that enables the identification of potential human-made threats, created by the usage of natural means and captured by heterogeneous sources (CCTV, UAV, etc.). Advanced machine learning techniques provide analysis of events and useful information, which are fused semantically and estimate the severity level of the potential attack, serving the needs for real-time monitoring and mitigating the risk.