scholarly journals Assessment of Convolutional Neural Network Architectures for Earthquake-Induced Building Damage Detection based on Pre- and Post-Event Orthophoto Images

2020 ◽  
Vol 12 (21) ◽  
pp. 3529
Author(s):  
Bahareh Kalantar ◽  
Naonori Ueda ◽  
Husam A. H. Al-Najjar ◽  
Alfian Abdul Halin
Keyword(s):  
Expert Knowledge ◽  
Building Damage ◽  
Network Architectures ◽  
Building Detection ◽  
Fusion Model ◽  
Twin Model ◽  
The 2016 Kumamoto Earthquake ◽  
Visual Technique ◽  
Damage Maps ◽  
Automatic Methods

In recent years, remote-sensing (RS) technologies have been used together with image processing and traditional techniques in various disaster-related works. Among these is detecting building damage from orthophoto imagery that was inflicted by earthquakes. Automatic and visual techniques are considered as typical methods to produce building damage maps using RS images. The visual technique, however, is time-consuming due to manual sampling. The automatic method is able to detect the damaged building by extracting the defect features. However, various design methods and widely changing real-world conditions, such as shadow and light changes, cause challenges to the extensive appointing of automatic methods. As a potential solution for such challenges, this research proposes the adaption of deep learning (DL), specifically convolutional neural networks (CNN), which has a high ability to learn features automatically, to identify damaged buildings from pre- and post-event RS imageries. Since RS data revolves around imagery, CNNs can arguably be most effective at automatically discovering relevant features, avoiding the need for feature engineering based on expert knowledge. In this work, we focus on RS imageries from orthophoto imageries for damaged-building detection, specifically for (i) background, (ii) no damage, (iii) minor damage, and (iv) debris classifications. The gist is to uncover the CNN architecture that will work best for this purpose. To this end, three CNN models, namely the twin model, fusion model, and composite model, are applied to the pre- and post-orthophoto imageries collected from the 2016 Kumamoto earthquake, Japan. The robustness of the models was evaluated using four evaluation metrics, namely overall accuracy (OA), producer accuracy (PA), user accuracy (UA), and F1 score. According to the obtained results, the twin model achieved higher accuracy (OA = 76.86%; F1 score = 0.761) compare to the fusion model (OA = 72.27%; F1 score = 0.714) and composite (OA = 69.24%; F1 score = 0.682) models.

scholarly journals Evolution of Deep Convolutional Neural Networks Using Cartesian Genetic Programming

2020 ◽  
Vol 28 (1) ◽  
pp. 141-163 ◽  
Author(s):  
Masanori Suganuma ◽  
Masayuki Kobayashi ◽  
Shinichi Shirakawa ◽  
Tomoharu Nagao
Keyword(s):  
Genetic Programming ◽  
Expert Knowledge ◽  
Computational Cost ◽  
Computational Time ◽  
Network Architectures ◽  
Cartesian Genetic Programming ◽  
Deep Convolutional Neural Networks ◽  
High Performing ◽  
Network Training ◽  
Automatic Design Methods

The convolutional neural network (CNN), one of the deep learning models, has demonstrated outstanding performance in a variety of computer vision tasks. However, as the network architectures become deeper and more complex, designing CNN architectures requires more expert knowledge and trial and error. In this article, we attempt to automatically construct high-performing CNN architectures for a given task. Our method uses Cartesian genetic programming (CGP) to encode the CNN architectures, adopting highly functional modules such as a convolutional block and tensor concatenation, as the node functions in CGP. The CNN structure and connectivity, represented by the CGP, are optimized to maximize accuracy using the evolutionary algorithm. We also introduce simple techniques to accelerate the architecture search: rich initialization and early network training termination. We evaluated our method on the CIFAR-10 and CIFAR-100 datasets, achieving competitive performance with state-of-the-art models. Remarkably, our method can find competitive architectures with a reasonable computational cost compared to other automatic design methods that require considerably more computational time and machine resources.

Investigation of Damages in Immediate Vicinity of Co-Seismic Faults During the 2016 Kumamoto Earthquake

2017 ◽  
Vol 12 (5) ◽  
pp. 899-915 ◽  
Author(s):  
Shohei Naito ◽  
Ken Xiansheng Hao ◽  
Shigeki Senna ◽  
Takuma Saeki ◽  
Hiromitsu Nakamura ◽  
...  
Keyword(s):  
Fault Zone ◽  
Active Fault ◽  
Strong Motion ◽  
Site Amplification ◽  
Building Damage ◽  
2016 Kumamoto Earthquake ◽  
Severe Damage ◽  
Kumamoto Earthquake ◽  
The 2016 Kumamoto Earthquake ◽  
The Relationship

In the 2016 Kumamoto earthquake, the Futagawa fault zone and the Hinagu fault zone were active in some sections, causing severe damage in neighboring areas along the faults. We conducted a detailed investigation of the surface earthquake fault, building damage, and site amplification of shallow ground within about 1 km of the neighboring areas of the fault. The focus was mainly on Kawayou district, Minamiaso village and Miyazono district, Mashiki town, and locations that suffered particularly severe building damage. We explored the relationship between local strong motion and building damage caused in areas that were in the immediate vicinity of the active fault.

scholarly journals Structural Building Damage Detection with Deep Learning: Assessment of a State-of-the-Art CNN in Operational Conditions

2019 ◽  
Vol 11 (23) ◽  
pp. 2765 ◽  
Author(s):  
Francesco Nex ◽  
Diogo Duarte ◽  
Fabio Giulio Tonolo ◽  
Norman Kerle
Keyword(s):  
Deep Learning ◽  
Damage Detection ◽  
Structural Damage ◽  
Relevant Information ◽  
Building Damage ◽  
Computational Time ◽  
Learning Networks ◽  
Remotely Sensed Data ◽  
Operational Conditions ◽  
Damage Maps

Remotely sensed data can provide the basis for timely and efficient building damage maps that are of fundamental importance to support the response activities following disaster events. However, the generation of these maps continues to be mainly based on the manual extraction of relevant information in operational frameworks. Considering the identification of visible structural damages caused by earthquakes and explosions, several recent works have shown that Convolutional Neural Networks (CNN) outperform traditional methods. However, the limited availability of publicly available image datasets depicting structural disaster damages, and the wide variety of sensors and spatial resolution used for these acquisitions (from space, aerial and UAV platforms), have limited the clarity of how these networks can effectively serve First Responder needs and emergency mapping service requirements. In this paper, an advanced CNN for visible structural damage detection is tested to shed some light on what deep learning networks can currently deliver, and its adoption in realistic operational conditions after earthquakes and explosions is critically discussed. The heterogeneous and large datasets collected by the authors covering different locations, spatial resolutions and platforms were used to assess the network performances in terms of transfer learning with specific regard to geographical transferability of the trained network to imagery acquired in different locations. The computational time needed to deliver these maps is also assessed. Results show that quality metrics are influenced by the composition of training samples used in the network. To promote their wider use, three pre-trained networks—optimized for satellite, airborne and UAV image spatial resolutions and viewing angles—are made freely available to the scientific community.

Towards a rapid automatic detection of building damage using remote sensing for disaster management

2014 ◽  
Vol 23 (1) ◽  
pp. 53-66 ◽  
Author(s):  
Thi-Thanh-Hiên Pham ◽  
Philippe Apparicio ◽  
Christopher Gomez ◽  
Christiane Weber ◽  
Dominique Mathon
Keyword(s):  
Disaster Management ◽  
Spatial Resolution ◽  
Rapid Assessment ◽  
Building Damage ◽  
Visual Interpretation ◽  
Content Type ◽  
Emergency Situations ◽  
Optical Images ◽  
Damage Maps ◽  
Damage Mapping

Purpose – Satellite and airborne images are increasingly used at different stages of disaster management, especially in the detection of infrastructure damage. Although semi- or full automatic techniques to detect damage have been proposed, they have not been used in emergency situations. Damage maps produced by international organisations are still based on visual interpretation of images, which is time- and labour-consuming. The purpose of this paper is to investigate how an automatic mapping of damage can be helpful for a first and rapid assessment of building damage. Design/methodology/approach – The study area is located in Port-au-Prince (Haiti) stricken by an earthquake in January 2010. To detect building damage, the paper uses optical images (15 cm of spatial resolution) coupled with height data (LiDAR, 1 m of spatial resolution). By undertaking an automatic object-oriented classification, the paper identifies three categories of building damages: intact buildings, collapsed buildings and debris. Findings – Data processing for the study area covering 11 km2 took about 15 hours. The accuracy of the classification varies from 70 to 79 per cent depending to the methods of assessment. Causes of errors are numerous: limited spectral information of the optical images, resolution difference between the two data, high density of buildings but most importantly, certain types of building collapses could not be detected by vertically taken images (the case of data in this study). Originality/value – The automatic damage mapping developed in this paper proves to be reliable and could be used in emergency situations. It could also be combined with manual visual interpretation to accelerate the planning of humanitarian rescues and reconstruction.

Building Damage Detection Based on Single-Phase High-Resolution Remote Sensing Images

2012 ◽  
Vol 433-440 ◽  
pp. 6422-6429
Author(s):  
Hong Zhang
Keyword(s):  
Damage Detection ◽  
Single Phase ◽  
Color Space ◽  
Detection Algorithm ◽  
Building Damage ◽  
Google Earth ◽  
Building Detection ◽  
Hsv Color Space ◽  
Detection Approach ◽  
Gray Level Histogram

This paper presents a building detection approach based on HSV color space. The method is based on the gray level histogram features, which can separate the housing construction units from complex background. A building damage detection algorithm based on regional statistical information is also proposed in this paper, and a set of performance parameters of feature vector is studied to identify the extent of the housing collapse. The experiments on Haiti post-earthquake images from Google Earth and Yushu post-earthquake images from Internet are discussed in the paper. The experimental results show that proposed approach is effective and feasible.

scholarly journals Method to Generate Building Damage Maps by Combining Aerial Image Processing and Crowdsourcing

2021 ◽  
Vol 16 (5) ◽  
pp. 827-839
Author(s):  
Hidehiko Shishido ◽  
◽  
Koyo Kobayashi ◽  
Yoshinari Kameda ◽  
Itaru Kitahara
Keyword(s):  
Image Processing ◽  
Large Scale ◽  
Information Source ◽  
Building Damage ◽  
Aerial Image ◽  
Information Collection ◽  
Essential Information ◽  
Disaster Information ◽  
Damage Maps ◽  
Damage Map

Building damage maps that show the damage status of buildings are an essential information source for various disaster countermeasures, such as evacuation, rescue, and reconstruction. Therefore, they must be generated as quickly as possible. However, to generate a building damage map, it is necessary to collect disaster information and estimate the damage situation over a wide area, which is time consuming. (In this paper, we consider disaster information collection as capturing aerial images.) In recent years, crowdsourcing has been widely used to understand the damage situation. Crowdsourcing achieves large-scale work by dividing it into microtasks that can be solved by anyone and by distributing the microtasks among an unspecified number of workers. We believe that crowdsourcing is suitable for gathering information and assessing damage situations as it can adjust the type and number of workers in a scalable manner and allocate resources according to the size of the disaster. Therefore, crowdsourcing has been used for gathering information and assessing the situation during disaster management. However, usually, the two types of crowdsourcing tasks (i.e., gathering information and assessing the damage) are performed independently; consequently, the collected information is often not utilized effectively. More efficient work can be expected by linking the two crowdsourcing tasks. This paper proposes a framework for efficiently generating a building damage map by combining the two methods of information collection on disaster areas and assessment of disaster situations using aerial image processing. The results of an experiment using a prototype of our proposed framework clarify the range of applications in the collection and assessment crowdsourcing tasks. The experimental results indicate the feasibility of understanding disaster situations using our method. In addition, it is possible to install artificial intelligence workers that can support human workers to estimate the damage situation more quickly.

The Importance of Seismic Death Risk Assessment of Households in the Kumamoto Earthquake of 2016

2017 ◽  
Vol 12 (6) ◽  
pp. 1151-1160 ◽  
Author(s):  
Tadayoshi Nakashima ◽  
◽  
Shigeyuki Okada ◽  
Akane Shinoda
Keyword(s):  
Risk Assessment ◽  
Decision Making ◽  
Building Damage ◽  
2016 Kumamoto Earthquake ◽  
Death Risk ◽  
Kumamoto Earthquake ◽  
The 2016 Kumamoto Earthquake ◽  
Return Home ◽  
Reduction Effect ◽  
Over Time

This paper discusses the reduction effect of a foreshock on casualties during the mainshock of people who evacuated to shelters and their private cars during the 2016 Kumamoto earthquake. In the first part of this paper, we discuss the number of human casualties caused by the collapse of wooden buildings. The characteristics of casualties in the Kumamoto earthquake are classified as household attributes and building damage caused by the foreshock and mainshock. In the second part, we apply equations (Nakashima and Okada 2008 and Okada and Nakashima 2015) to the Masiki area to determine the total number of casualties with a focus on deaths. The number of deaths due to total building damage from the foreshock and the mainshock in the case of 0 evacuees was estimated as 147. We then estimated the reduction effect on the number of casualties caused by the foreshock by using the survey data of the mainshock and foreshock. We found that evacuation during the mainshock decreased the death toll by 128 people. Moreover, the number of injured people decreased by 657. Generally, most people who evacuate tend to return home over time. As a result, many people die at the time of a subsequent mainshock. It is important to provide death risk information to each household to support their decision-making regarding appropriate evacuation.

scholarly journals Detection of Earthquake-Induced Building Damages Using Polarimetric SAR Data

2020 ◽  
Vol 12 (1) ◽  
pp. 137 ◽  
Author(s):  
Sang-Eun Park ◽  
Yoon Taek Jung
Keyword(s):  
Change Detection ◽  
Urban Areas ◽  
Detection Method ◽  
Building Damage ◽  
Polarimetric Sar ◽  
Ground Shaking ◽  
The 2016 Kumamoto Earthquake ◽  
Radar Scattering ◽  
The Matrix ◽  
Damage Indicators

Remote sensing, particularly using synthetic aperture radar (SAR) systems, can be an effective tool in detecting and assessing the area and amount of building damages caused by earthquake or tsunami. Several studies have provided experimental evidence for the importance of polarimetric SAR observations in building damage detection and assessment, particularly caused by a tsunami. This study aims to evaluate the practical applicability of the polarimetric SAR observations to building damage caused by the direct ground-shaking of an earthquake. The urban areas heavily damaged by the 2016 Kumamoto earthquake in Japan have been investigated by using the polarimetric PALSAR-2 data acquired in pre- and post-earthquake conditions. Several polarimetric change detection approaches, such as the changes of polarimetric scattering powers, the matrix dissimilarity measures, and changes of the radar scattering mechanisms, were examined. Optimal damage indicators in the presence of significant natural changes, and a novel change detection method by the fuzzy-based fusion of polarimetric damage indicators are proposed. The accuracy analysis results show that the proposed automatic classification method can successfully detect the selected damaged areas with a detection rate of 90.9% and false-alarm rate of 1.3%.

Sign in / Sign up

Export Citation Format

Share Document