Self-Incremental Learning for Rapid Identification of Collapsed Buildings Triggered by Natural Disasters
The building damage caused by natural disasters seriously threatens human security. Applying deep learning algorithms to identify collapsed buildings from remote sensing images is crucial for rapid post-disaster emergency response. However, the diversity of buildings, limited training dataset size,...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2023-08-01
|
Series: | Remote Sensing |
Subjects: | |
Online Access: | https://www.mdpi.com/2072-4292/15/15/3909 |
_version_ | 1797586039375659008 |
---|---|
author | Jiayi Ge Hong Tang Chao Ji |
author_facet | Jiayi Ge Hong Tang Chao Ji |
author_sort | Jiayi Ge |
collection | DOAJ |
description | The building damage caused by natural disasters seriously threatens human security. Applying deep learning algorithms to identify collapsed buildings from remote sensing images is crucial for rapid post-disaster emergency response. However, the diversity of buildings, limited training dataset size, and lack of ground-truth samples after sudden disasters can significantly reduce the generalization of a pre-trained model for building damage identification when applied directly to non-preset locations. To address this challenge, a self-incremental learning framework (i.e., SELF) is proposed in this paper, which can quickly improve the generalization ability of the pre-trained model in disaster areas by self-training an incremental model using automatically selected samples from post-disaster images. The effectiveness of the proposed method is verified on the 2010 Yushu earthquake, 2023 Turkey earthquake, and other disaster types. The experimental results demonstrate that our approach outperforms state-of-the-art methods in terms of collapsed building identification, with an average increase of more than 6.4% in the Kappa coefficient. Furthermore, the entire process of the self-incremental learning method, including sample selection, incremental learning, and collapsed building identification, can be completed within 6 h after obtaining the post-disaster images. Therefore, the proposed method is effective for emergency response to natural disasters, which can quickly improve the application effect of the deep learning model to provide more accurate building damage results. |
first_indexed | 2024-03-11T00:17:36Z |
format | Article |
id | doaj.art-9881c21e30ff463c96af3159175942e2 |
institution | Directory Open Access Journal |
issn | 2072-4292 |
language | English |
last_indexed | 2024-03-11T00:17:36Z |
publishDate | 2023-08-01 |
publisher | MDPI AG |
record_format | Article |
series | Remote Sensing |
spelling | doaj.art-9881c21e30ff463c96af3159175942e22023-11-18T23:32:28ZengMDPI AGRemote Sensing2072-42922023-08-011515390910.3390/rs15153909Self-Incremental Learning for Rapid Identification of Collapsed Buildings Triggered by Natural DisastersJiayi Ge0Hong Tang1Chao Ji2State Key Laboratory of Remote Sensing Science, Faculty of Geographical Science, Beijing Normal University, Beijing 100875, ChinaState Key Laboratory of Remote Sensing Science, Faculty of Geographical Science, Beijing Normal University, Beijing 100875, ChinaState Key Laboratory of Remote Sensing Science, Faculty of Geographical Science, Beijing Normal University, Beijing 100875, ChinaThe building damage caused by natural disasters seriously threatens human security. Applying deep learning algorithms to identify collapsed buildings from remote sensing images is crucial for rapid post-disaster emergency response. However, the diversity of buildings, limited training dataset size, and lack of ground-truth samples after sudden disasters can significantly reduce the generalization of a pre-trained model for building damage identification when applied directly to non-preset locations. To address this challenge, a self-incremental learning framework (i.e., SELF) is proposed in this paper, which can quickly improve the generalization ability of the pre-trained model in disaster areas by self-training an incremental model using automatically selected samples from post-disaster images. The effectiveness of the proposed method is verified on the 2010 Yushu earthquake, 2023 Turkey earthquake, and other disaster types. The experimental results demonstrate that our approach outperforms state-of-the-art methods in terms of collapsed building identification, with an average increase of more than 6.4% in the Kappa coefficient. Furthermore, the entire process of the self-incremental learning method, including sample selection, incremental learning, and collapsed building identification, can be completed within 6 h after obtaining the post-disaster images. Therefore, the proposed method is effective for emergency response to natural disasters, which can quickly improve the application effect of the deep learning model to provide more accurate building damage results.https://www.mdpi.com/2072-4292/15/15/3909building damageremote sensingself-incremental learningsample selectiondisaster emergency response |
spellingShingle | Jiayi Ge Hong Tang Chao Ji Self-Incremental Learning for Rapid Identification of Collapsed Buildings Triggered by Natural Disasters Remote Sensing building damage remote sensing self-incremental learning sample selection disaster emergency response |
title | Self-Incremental Learning for Rapid Identification of Collapsed Buildings Triggered by Natural Disasters |
title_full | Self-Incremental Learning for Rapid Identification of Collapsed Buildings Triggered by Natural Disasters |
title_fullStr | Self-Incremental Learning for Rapid Identification of Collapsed Buildings Triggered by Natural Disasters |
title_full_unstemmed | Self-Incremental Learning for Rapid Identification of Collapsed Buildings Triggered by Natural Disasters |
title_short | Self-Incremental Learning for Rapid Identification of Collapsed Buildings Triggered by Natural Disasters |
title_sort | self incremental learning for rapid identification of collapsed buildings triggered by natural disasters |
topic | building damage remote sensing self-incremental learning sample selection disaster emergency response |
url | https://www.mdpi.com/2072-4292/15/15/3909 |
work_keys_str_mv | AT jiayige selfincrementallearningforrapididentificationofcollapsedbuildingstriggeredbynaturaldisasters AT hongtang selfincrementallearningforrapididentificationofcollapsedbuildingstriggeredbynaturaldisasters AT chaoji selfincrementallearningforrapididentificationofcollapsedbuildingstriggeredbynaturaldisasters |