Damaged Building Recognition of the Earthquakes Using Deep Learning with Field and Aerial Photographs
| Project/Area Number |
16K12834
|
|---|
| Research Category |
Grant-in-Aid for Challenging Exploratory Research
|
| Allocation Type | Multi-year Fund |
|---|
| Research Field |
Social systems engineering/Safety system
|
|---|
| Research Institution | Tokyo Institute of Technology |
|---|
Principal Investigator |
Matsuoka Masashi 東京工業大学, 環境・社会理工学院, 准教授 (80242311)
|
|---|
| Research Collaborator |
MAKI Norio 京都大学, 防災研究所, 教授
TANAKA Satoshi 常葉大学, 大学院環境防災研究科, 教授
NAKAMURA Ryosuke 産業技術総合研究所, 人工知能研究センター, チーム長
|
|---|
| Project Period (FY) |
2016-04-01 – 2018-03-31
|
| Project Status |
Completed (Fiscal Year 2017)
|
| Budget Amount *help |
¥3,510,000 (Direct Cost: ¥2,700,000、Indirect Cost: ¥810,000)
Fiscal Year 2017: ¥2,080,000 (Direct Cost: ¥1,600,000、Indirect Cost: ¥480,000)
Fiscal Year 2016: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
|
|---|
| Keywords | 地震建物被害 / 被災度判定 / 深層学習 / 航空写真 / 現地写真 / 地震被害 / 建物 / CNN / 2016年熊本地震 / ディープラーニング / 阪神・淡路大震災 / 防災 / 建物被災度 / 機械学習 / 写真 |
|---|
| Outline of Final Research Achievements |
We conducted Convolutional Neural Network (CNN) for the 1995 Kobe earthquake and the 2016 Kumamoto earthquakes to learn images of damaged buildings, generated classifiers for building damage, estimated the building damage and compared with the field survey data. As a result from the Kobe dataset, it became clear that buildings with severe damage can be identified with accuracy of 86.0% for aerial photographs and 83.0% for field photographs. In addition, from the photograph of the site it became clear that the collapsed building can be identified with an accuracy of 98.5%. Also, for the purpose of supplementing the point that not all damage parts can be judged from aerial photographs, estimated seismic intensity at the location of the building was added to the input of learning of CNN. As a result from the Kumamoto dataset, it was shown that there is a possibility of detailed estimation of building damage degree based on the severely damage probability output from CNN.
|
Report
(3 results)
Research Products
(10 results)
