Please use this identifier to cite or link to this item: https://bura.brunel.ac.uk/handle/2438/34193
Title: Satellite-Based Conflict Damage Detection: Siamese CNNs for Forced Displacement Planning in Ukraine
Authors: Matarova, Taulant
Suleimenova, Diana
Keywords: Siamese CNN;damage detection;satellite imagery;population displacement
Issue Date: 29-Jun-2026
Publisher: Springer Nature Switzerland
Citation: Matarova, T. and Suleimenova, D. (2026) 'Satellite-Based Conflict Damage Detection: Siamese CNNs for Forced Displacement Planning in Ukraine', in M. Paszynski, A.S. Barnard and Y.J. Zhang (eds.) Computational Science – ICCS 2026 Workshops (Lecture Notes in Computer Science ). Cham: Springer Nature Switzerland, pp. 58–72. doi: 10.1007/978-3-032-29915-4_5.
Series/Report no.: Lecture Notes in Computer Science;LNCS 16788
Abstract: Forced displacement due to armed conflict is an escalating global challenge, with over 117 million people displaced worldwide in 2025. Traditional damage assessment methods are slow, resource-intensive, and often impractical in active conflict zones. We present a Siamese Convolutional Neural Network for binary building damage detection using freely available Sentinel-2 medium-resolution satellite imagery. Using pre- and post-conflict image pairs from the Copernicus Data Space Ecosystem and UNOSAT building damage annotations, we evaluate eight model variants across four ResNet backbone depths with a Multilayer Perceptron classification head, tested across twenty-two conflict-affected Ukrainian regions. Our best-performing model, a Siamese ResNet-101 with an MLP head, achieves an Area Under the ROC Curve of 0.911 and an Average Precision of 0.888, demonstrating strong detection capability without high-resolution commercial imagery. We analyse the model’s utility for both outward and inward humanitarian displacement planning and discuss its potential integration with agent-based models for simulating forced displacement scenarios. Our results demonstrate that resource-efficient, publicly available satellite data can support data-driven humanitarian planning at scale.
Description: Data Availability Statement: The code for all eight model variants and hardware specifications are available at https://github.com/TMatarova/S2_urban_damage_detector_siamese. The dataset and pretrained weights for the best-performing model are additionally archived on Zenodo at https://zenodo.org/records/19441105.
URI: https://bura.brunel.ac.uk/handle/2438/34193
DOI: https://doi.org/10.1007/978-3-032-29915-4_5
ISBN: 9783032299147
9783032299154
ISSN: 0302-9743
Appears in Collections:Department of Computer Science Embargoed Research Papers

Files in This Item:
File Description SizeFormat 
FullText.pdfEmbargoed until 29 June 2027. Copyright © 2026 The Author(s), under exclusive license to Springer Nature Switzerland AG. This is a pre-copyedited, author-produced version of a book chapter accepted for publication In: Paszynski, M., Barnard, A.S., Zhang, Y.J. (eds) Computational Science – ICCS 2026 Workshops. Lecture Notes in Computer Science, vol 1678, following peer review. The final authenticated version is available1.67 MBAdobe PDFView/Open


Items in BURA are protected by copyright, with all rights reserved, unless otherwise indicated.