dc.contributor.author |
Myntiuk, Sofiia
|
|
dc.date.accessioned |
2024-02-14T11:33:29Z |
|
dc.date.available |
2024-02-14T11:33:29Z |
|
dc.date.issued |
2023 |
|
dc.identifier.citation |
Myntiuk, Sofiia. Identifying the Effects of Russian Aggression on Agricultural Fields in
Ukraine through Classification Approaches and Satellite Imagery / Sofiia Myntiuk; Supervisor: Oles Dobosevych; Ukrainian Catholic University, Department of Computer Sciences. – Lviv: 2023. – 42 p.: ill. |
uk |
dc.identifier.uri |
https://er.ucu.edu.ua/handle/1/4427 |
|
dc.language.iso |
en |
uk |
dc.title |
Identifying the Effects of Russian Aggression on Agricultural Fields in Ukraine through Classification Approaches and Satellite Imagery |
uk |
dc.type |
Preprint |
uk |
dc.status |
Опублікований і розповсюджений раніше |
uk |
dc.description.abstracten |
Detection and assessment of shelling-induced damage to the agricultural fields of
Ukraine are crucial for ensuring the safety of civilians, as it helps to estimate the
number of unexploded ordnances in the region. Most existing approaches to dam-
age detection solve this problem for buildings, and the crater detection task is usu-
ally solved either for historical or planetary images. In this thesis, we aim to explore
the applicability of existing approaches to the task of crater detection in agricul-
tural fields in Ukraine. We collect and annotate a dataset with satellite images of
Ukrainian agricultural fields. We experiment with solutions that include classification methods and conduct the hyperparameter search to find the best model for our
data. We analyze the impact of each hyperparameter on the network performance
and demonstrate the network’s ability to generalize to new locations.
The code used in our solution can be found in the project’s GitHub repository. |
uk |