Article
Journal/Revue :
International Journal of Image and Data Fusion
ISSN : 1947-9832
Publisher :
Taylor \& Francis
Informations
Année : 2022
Volume : 13 Numéro : 1
Pages : 1-20
Détails
Improving damage classification via hybrid deep learning feature representations derived from post-earthquake aerial images
Tarablesse Settou Mohamed-Khireddine Kholladi Abdelkamel Ben Ali
One of the crucial problems after earthquakes is how to quickly and accurately detect and identify damaged areas. Several automated methods have been developed to analyse remote sensing (RS) images for earthquake damage classification. The performance of damage classification is mainly depending on powerful learning feature representations. Though the hand-crafted features can achieve satisfactory performance to some extent, the performance gain is small and does not generalise well. Recently, the convolutional neural network (CNN) has demonstrated its capability of deriving more powerful feature representations than hand-crafted features in many domains. Our main contribution in this paper is the investigation of hybrid feature representations derived from several pre-trained CNN models for earthquake damage classification. Also, in this study, in contrast to previous works, we explore the combination of feature representations extracted from the last two fully connected layers of a particular CNN model. We validated our proposals on two large datasets, including images highly varying in scene characteristics, lighting conditions, and image characteristics, captured from different earthquake events and several geographic locations. Extensive experiments showed that our proposals can improve significantly the performance.
Mots clés :
Remote sensing Earthquake damage classification Hybrid feature representation Pre-trained CNN Deep learning
Réf. de citation :
misc-lab-400
DOI :
10.1080/19479832.2020.1864787
Lien :
Texte intégral
ACM :
T. Settou, M.-K. Kholladi and A. B. Ali. 2022. Improving damage classification via hybrid deep learning feature representations derived from post-earthquake aerial images. International Journal of Image and Data Fusion, 13, 1, Taylor \& Francis, 1-20. DOI: https://doi.org/10.1080/19479832.2020.1864787.
APA :
Settou, T., Kholladi, M.-K. & Ali, A. B. (2022). Improving damage classification via hybrid deep learning feature representations derived from post-earthquake aerial images. International Journal of Image and Data Fusion, 13(1), Taylor \& Francis, 1-20. DOI: https://doi.org/10.1080/19479832.2020.1864787
IEEE :
T. Settou, M.-K. Kholladi and A. B. Ali, "Improving damage classification via hybrid deep learning feature representations derived from post-earthquake aerial images". International Journal of Image and Data Fusion, vol. 13, no. 1, Taylor \& Francis, pp. 1-20, 2022. DOI: https://doi.org/10.1080/19479832.2020.1864787.
BibTeX :
@article{misc-lab-400,
author = {Settou, Tarablesse and Kholladi, Mohamed-Khireddine and Ali, Abdelkamel Ben},
title = {Improving damage classification via hybrid deep learning feature representations derived from post-earthquake aerial images},
journal = {International Journal of Image and Data Fusion},
volume = {13},
number = {1},
issn = {1947-9832},
pages = {1--20},
publisher = {Taylor \& Francis},
year = {2022},
doi = {10.1080/19479832.2020.1864787},
url = {https://www.tandfonline.com/doi/abs/10.1080/19479832.2020.1864787},
keywords = {Remote sensing, earthquake damage classification, hybrid feature representation, pre-trained CNN, deep learning}
}
RIS :
TI  - Improving damage classification via hybrid deep learning feature representations derived from post-earthquake aerial images
AU - T. Settou
AU - M.-K. Kholladi
AU - A. B. Ali
PY - 2022
SN - 1947-9832
JO - International Journal of Image and Data Fusion
VL - 13
IS - 1
SP - 1
EP - 20
PB - Taylor \& Francis
AB - One of the crucial problems after earthquakes is how to quickly and accurately detect and identify damaged areas. Several automated methods have been developed to analyse remote sensing (RS) images for earthquake damage classification. The performance of damage classification is mainly depending on powerful learning feature representations. Though the hand-crafted features can achieve satisfactory performance to some extent, the performance gain is small and does not generalise well. Recently, the convolutional neural network (CNN) has demonstrated its capability of deriving more powerful feature representations than hand-crafted features in many domains. Our main contribution in this paper is the investigation of hybrid feature representations derived from several pre-trained CNN models for earthquake damage classification. Also, in this study, in contrast to previous works, we explore the combination of feature representations extracted from the last two fully connected layers of a particular CNN model. We validated our proposals on two large datasets, including images highly varying in scene characteristics, lighting conditions, and image characteristics, captured from different earthquake events and several geographic locations. Extensive experiments showed that our proposals can improve significantly the performance.
KW - Remote sensing
KW - earthquake damage classification
KW - hybrid feature representation
KW - pre-trained CNN
KW - deep learning
DO - 10.1080/19479832.2020.1864787
UR - https://www.tandfonline.com/doi/abs/10.1080/19479832.2020.1864787
ID - misc-lab-400
ER -