Our faculty member, Assoc. Prof. Gülşen Taşkın, was involved in a recent study proposing a deep learning-based classifier for automated post-earthquake damage assessment using images of buildings.

The study utilizes ConvNeXt, a state-of-the-art convolutional neural network, to distinguish between structural and nonstructural damage. The model was fine-tuned with transfer learning on a dataset of 9,645 labeled images from buildings affected by the 2020 Elazığ Earthquake, and later tested on data from the 2023 Kahramanmaraş Earthquake.

With advanced techniques such as data augmentation and regularization, the model achieved strong generalization and high classification accuracy, showing great promise for reliable post-disaster assessments.

To read the full article: springer.com/article/10.1007/s13369-025-10279-7