| A DEEP LEARNING FRAMEWORK FOR STREETS DAMAGE ASSESSMENT USING POST-EARTHQUAKE LIDAR DATA |
| Paper ID : 1329-SMPR |
| Authors: |
|
Seyd Teymoor Seydi1, Heidar Rastiveis *2 1Photogrametry and Remote Sensing Department, School of Surveying and Geospatial Engineering, College of Engineering, University of Tehran, Tehran, Iran 2Department of Photogrammetry and Remote Sensing,
School of Surveying and Geo-Spatial Engineering
College of Engineering, University of Tehran, |
| Abstract: |
| Streets network are the most important parts of urban infrastructures which can cause disorder to the city whenever they undergo a problem. This paper aims to provide and implement a deep-feature based method to determine streets network situation after an earthquake using LiDAR point cloud. The proposed framework works in three main phases: in the first phase, deep features of LiDAR data are extracted using a Convolutional Neural Network (CNN). Then, applying the extracted features in a multilayer perceptron (MLP) neural network, debris areas inside the streets network are detected. Eventually, the amount of debris in each street is applied for determining the blocked or un-blocked roads. To evaluate the efficiency of the proposed framework, LiDAR point cloud of the Port-au-Prince, Haiti after the 2010 Haiti earthquake was applied. The overall accuracy of more than 96% and kappa coefficient of 0.94 proved the high performance of this framework for debris detection. Moreover, analyzing damage assessment of 25 roads and comparing to a visually generated damaged map, 22 of the roads were correctly estimated in blocked or un-blocked classes. |
| Keywords: |
| LiDAR, Earthquake, Deep learning, Conventional Neural Network, Roads damage map, Haiti Earthquake |
| Status : Conditional Accept (Oral Presentation) |
