A carregar...
Publicação
Deep Learning-Based Documentation of Conflict Destroyed Buildings in Myanmar with VHR Satellite Images
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 9.59 MB | Adobe PDF |
Autores
Resumo(s)
In the wake of the 2021 military coup in Myanmar, the nation has been subjected to severe human rights abuses, prominently featuring arson attacks by the military on villages exhibiting resistance. The challenge of documenting these attacks at the individual building level is substantial, necessitating an approach that surpasses the limitations of conventional methods in terms of precision, scalability, and speed to support accountability and transitional justice. This thesis introduces an automated workflow that leverages bitemporal high and very high-resolution satellite imagery, combining Mask R-CNN for instance-based building footprint extraction with spatial overlap analysis and a Siamese Network for change detection. This methodology not only identifies destroyed structures with high accuracy but also provides an interface for potentially linking documented damage to individual victims through auxiliary data such as cadastral databases in future research endeavors. By providing a method that integrates advanced deep learning techniques for change detection at the building level, this work contributes a potential tool for destruction documentation by different entities and actors, enhancing the efficiency, granularity and accuracy of their documentation work. The implications of this work improving the potential for accountability and aiding transitional justice processes by proving an efficient mean for documenting human rights violations in conflict-affected regions.
Descrição
Dissertation submitted in partial fulfilment of the requirements for the Degree of Master of Science in Geospatial Technologies
Palavras-chave
Conflict Destruction Change Detection Deep Learning Remote Sensing Transitional Justice SDG 11 - Sustainable cities and communities SDG 13 - Climate action SDG 16 - Peace, justice and strong institutions