BDD-Net: an end-to-end multiscale residual CNN for earthquake-induced building damage detection
Building damage maps can be generated from either optical or Light Detection and Ranging (Lidar) datasets. In the wake of a disaster such as an earthquake, a timely and detailed map is a critical reference for disaster teams in order to plan and perform rescue and evacuation missions. Recent studies...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Published: |
MDPI
2022
|
| Online Access: | http://psasir.upm.edu.my/id/eprint/100516/ |
| _version_ | 1848863344182689792 |
|---|---|
| author | Seydi, Seyd Teymoor Rastiveis, Heidar Kalantar, Bahareh Abdul Halin, Alfian Ueda, Naonori |
| author_facet | Seydi, Seyd Teymoor Rastiveis, Heidar Kalantar, Bahareh Abdul Halin, Alfian Ueda, Naonori |
| author_sort | Seydi, Seyd Teymoor |
| building | UPM Institutional Repository |
| collection | Online Access |
| description | Building damage maps can be generated from either optical or Light Detection and Ranging (Lidar) datasets. In the wake of a disaster such as an earthquake, a timely and detailed map is a critical reference for disaster teams in order to plan and perform rescue and evacuation missions. Recent studies have shown that, instead of being used individually, optical and Lidar data can potentially be fused to obtain greater detail. In this study, we explore this fusion potential, which incorporates deep learning. The overall framework involves a novel End-to-End convolutional neural network (CNN) that performs building damage detection. Specifically, our building damage detection network (BDD-Net) utilizes three deep feature streams (through a multi-scale residual depth-wise convolution block) that are fused at different levels of the network. This is unlike other fusion networks that only perform fusion at the first and the last levels. The performance of BDD-Net is evaluated under three different phases, using optical and Lidar datasets for the 2010 Haiti Earthquake. The three main phases are: (1) data preprocessing and building footprint extraction based on building vector maps, (2) sample data preparation and data augmentation, and (3) model optimization and building damage map generation. The results of building damage detection in two scenarios show that fusing the optical and Lidar datasets significantly improves building damage map generation, with an overall accuracy (OA) greater than 88%. |
| first_indexed | 2025-11-15T13:31:25Z |
| format | Article |
| id | upm-100516 |
| institution | Universiti Putra Malaysia |
| institution_category | Local University |
| last_indexed | 2025-11-15T13:31:25Z |
| publishDate | 2022 |
| publisher | MDPI |
| recordtype | eprints |
| repository_type | Digital Repository |
| spelling | upm-1005162023-11-21T08:30:33Z http://psasir.upm.edu.my/id/eprint/100516/ BDD-Net: an end-to-end multiscale residual CNN for earthquake-induced building damage detection Seydi, Seyd Teymoor Rastiveis, Heidar Kalantar, Bahareh Abdul Halin, Alfian Ueda, Naonori Building damage maps can be generated from either optical or Light Detection and Ranging (Lidar) datasets. In the wake of a disaster such as an earthquake, a timely and detailed map is a critical reference for disaster teams in order to plan and perform rescue and evacuation missions. Recent studies have shown that, instead of being used individually, optical and Lidar data can potentially be fused to obtain greater detail. In this study, we explore this fusion potential, which incorporates deep learning. The overall framework involves a novel End-to-End convolutional neural network (CNN) that performs building damage detection. Specifically, our building damage detection network (BDD-Net) utilizes three deep feature streams (through a multi-scale residual depth-wise convolution block) that are fused at different levels of the network. This is unlike other fusion networks that only perform fusion at the first and the last levels. The performance of BDD-Net is evaluated under three different phases, using optical and Lidar datasets for the 2010 Haiti Earthquake. The three main phases are: (1) data preprocessing and building footprint extraction based on building vector maps, (2) sample data preparation and data augmentation, and (3) model optimization and building damage map generation. The results of building damage detection in two scenarios show that fusing the optical and Lidar datasets significantly improves building damage map generation, with an overall accuracy (OA) greater than 88%. MDPI 2022-05-05 Article PeerReviewed Seydi, Seyd Teymoor and Rastiveis, Heidar and Kalantar, Bahareh and Abdul Halin, Alfian and Ueda, Naonori (2022) BDD-Net: an end-to-end multiscale residual CNN for earthquake-induced building damage detection. Remote Sens, 14 (9). art. no. 2214. pp. 1-20. ISSN 2072-4292 https://www.mdpi.com/2072-4292/14/9/2214 10.3390/rs14092214 |
| spellingShingle | Seydi, Seyd Teymoor Rastiveis, Heidar Kalantar, Bahareh Abdul Halin, Alfian Ueda, Naonori BDD-Net: an end-to-end multiscale residual CNN for earthquake-induced building damage detection |
| title | BDD-Net: an end-to-end multiscale residual CNN for earthquake-induced building damage detection |
| title_full | BDD-Net: an end-to-end multiscale residual CNN for earthquake-induced building damage detection |
| title_fullStr | BDD-Net: an end-to-end multiscale residual CNN for earthquake-induced building damage detection |
| title_full_unstemmed | BDD-Net: an end-to-end multiscale residual CNN for earthquake-induced building damage detection |
| title_short | BDD-Net: an end-to-end multiscale residual CNN for earthquake-induced building damage detection |
| title_sort | bdd-net: an end-to-end multiscale residual cnn for earthquake-induced building damage detection |
| url | http://psasir.upm.edu.my/id/eprint/100516/ http://psasir.upm.edu.my/id/eprint/100516/ http://psasir.upm.edu.my/id/eprint/100516/ |