Semisupervised Cross Domain Teacher–Student Mutual Training for Damaged Building Detection
Detection of damaged buildings is a form of object detection and is essential for disaster emergency response efforts. In recent years, deep learning has been widely used in object detection, with successful target detection models such as Faster-Rcnn and You Only Look Once (YOLO) being proposed. Ho...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2023-01-01
|
Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/10175376/ |
_version_ | 1797330410387013632 |
---|---|
author | Jie Pan Pengyu Yin Xian Sun Junxiang Tan Wei Li |
author_facet | Jie Pan Pengyu Yin Xian Sun Junxiang Tan Wei Li |
author_sort | Jie Pan |
collection | DOAJ |
description | Detection of damaged buildings is a form of object detection and is essential for disaster emergency response efforts. In recent years, deep learning has been widely used in object detection, with successful target detection models such as Faster-Rcnn and You Only Look Once (YOLO) being proposed. However, training deep learning models usually requires a large amount of labeled data. Due to the high threshold for aerial remote sensing data collection, labeled aerial data of collapsed buildings is very sparse. In addition, the limited area of damage in a single scene leads to insufficient feature diversity, which can easily lead to model overfitting. These issues restrict the development of deep learning in emergency response applications. To solve these problems, we propose a paradigm named cross-domain teacher–student mutual training. By using the Cycle-GAN-generated style transfer data through teacher network, pseudolabels are generated to train the student network. Then, the student network slowly updates the parameters of the teacher network to indirectly learn the generalization information of the satellite data domain. Networks trained in this way can achieve good results in detecting collapsed houses in aviation and satellite data. We tested the results on our self-built dataset, DB-ARSD, which includes bounding box labeling of the damaged buildings, and found that our method outperforms other object detection methods in both collapsed house prediction accuracy and domain transfer generalization performance. |
first_indexed | 2024-03-08T07:19:26Z |
format | Article |
id | doaj.art-81f7e16715cd467fa410ece3f1834e5e |
institution | Directory Open Access Journal |
issn | 2151-1535 |
language | English |
last_indexed | 2024-03-08T07:19:26Z |
publishDate | 2023-01-01 |
publisher | IEEE |
record_format | Article |
series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
spelling | doaj.art-81f7e16715cd467fa410ece3f1834e5e2024-02-03T00:01:08ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352023-01-01168191820310.1109/JSTARS.2023.329339710175376Semisupervised Cross Domain Teacher–Student Mutual Training for Damaged Building DetectionJie Pan0Pengyu Yin1https://orcid.org/0000-0001-7979-0436Xian Sun2https://orcid.org/0000-0002-0038-9816Junxiang Tan3https://orcid.org/0000-0002-8405-4852Wei Li4https://orcid.org/0000-0001-7015-7335Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing, ChinaAerospace Information Research Institute, Chinese Academy of Sciences, Beijing, ChinaAerospace Information Research Institute, Chinese Academy of Sciences, Beijing, ChinaAerospace Information Research Institute, Chinese Academy of Sciences, Beijing, ChinaSchool of Information and Electronics, Beijing Institute of Technology, Beijing, ChinaDetection of damaged buildings is a form of object detection and is essential for disaster emergency response efforts. In recent years, deep learning has been widely used in object detection, with successful target detection models such as Faster-Rcnn and You Only Look Once (YOLO) being proposed. However, training deep learning models usually requires a large amount of labeled data. Due to the high threshold for aerial remote sensing data collection, labeled aerial data of collapsed buildings is very sparse. In addition, the limited area of damage in a single scene leads to insufficient feature diversity, which can easily lead to model overfitting. These issues restrict the development of deep learning in emergency response applications. To solve these problems, we propose a paradigm named cross-domain teacher–student mutual training. By using the Cycle-GAN-generated style transfer data through teacher network, pseudolabels are generated to train the student network. Then, the student network slowly updates the parameters of the teacher network to indirectly learn the generalization information of the satellite data domain. Networks trained in this way can achieve good results in detecting collapsed houses in aviation and satellite data. We tested the results on our self-built dataset, DB-ARSD, which includes bounding box labeling of the damaged buildings, and found that our method outperforms other object detection methods in both collapsed house prediction accuracy and domain transfer generalization performance.https://ieeexplore.ieee.org/document/10175376/Aerial remote sensing datasetdamaged building detectiondeep learningdomain adaptionsemisupervised object detection |
spellingShingle | Jie Pan Pengyu Yin Xian Sun Junxiang Tan Wei Li Semisupervised Cross Domain Teacher–Student Mutual Training for Damaged Building Detection IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Aerial remote sensing dataset damaged building detection deep learning domain adaption semisupervised object detection |
title | Semisupervised Cross Domain Teacher–Student Mutual Training for Damaged Building Detection |
title_full | Semisupervised Cross Domain Teacher–Student Mutual Training for Damaged Building Detection |
title_fullStr | Semisupervised Cross Domain Teacher–Student Mutual Training for Damaged Building Detection |
title_full_unstemmed | Semisupervised Cross Domain Teacher–Student Mutual Training for Damaged Building Detection |
title_short | Semisupervised Cross Domain Teacher–Student Mutual Training for Damaged Building Detection |
title_sort | semisupervised cross domain teacher x2013 student mutual training for damaged building detection |
topic | Aerial remote sensing dataset damaged building detection deep learning domain adaption semisupervised object detection |
url | https://ieeexplore.ieee.org/document/10175376/ |
work_keys_str_mv | AT jiepan semisupervisedcrossdomainteacherx2013studentmutualtrainingfordamagedbuildingdetection AT pengyuyin semisupervisedcrossdomainteacherx2013studentmutualtrainingfordamagedbuildingdetection AT xiansun semisupervisedcrossdomainteacherx2013studentmutualtrainingfordamagedbuildingdetection AT junxiangtan semisupervisedcrossdomainteacherx2013studentmutualtrainingfordamagedbuildingdetection AT weili semisupervisedcrossdomainteacherx2013studentmutualtrainingfordamagedbuildingdetection |