Monitoring War Destruction

The Monitoring War Destruction (MONWADES) project aims to develop an automated system for assessing war-related destruction using high-resolution satellite imagery and machine learning. Current methods, which rely on manual detection and eyewitness reports, are often slow, expensive, and prone to bias. By integrating computer vision and deep learning techniques, MONWADES seeks to provide a more efficient, accurate, and cost-effective solution for damage assessment, initially focusing on Ukraine and later expanding to other conflict zones.
This initiative builds on the methodological foundations of the ERC project RUSTDEC and previous research conducted in Syria. The goal is to create a global war damage assessment model capable of adapting to different geographic regions, a challenge known as spatial domain shift. To achieve this, MONWADES will collaborate with UNOSAT (United Nations Satellite Centre), a key institution in geospatial analysis for humanitarian aid, ensuring that the model is practical and widely applicable.
The project involves designing a novel neural network model that efficiently leverages panel data to enhance damage detection
accuracy, expanding training datasets with destruction annotations from multiple conflict zones, and validating model performance through real-world use cases. Ultimately, MONWADES aims to deploy the system at UNOSAT, helping to reduce costs and improve resource allocation for humanitarian efforts. If successful, the project could transform post-war recovery initiatives, strengthen humanitarian responses, and contribute to accountability by providing reliable and objective destruction data.