Automated Image Classification for Post-Earthquake Reconnaissance Images

  • Juan Park
  • Chul Min Yeum
  • Jongseong Choi
  • Xiaoyu Liu

Abstract

In the aftermath of earthquake events, many reconnaissance
teams are dispatched to collect as much data as possible, moving
quickly to capture the damages and failures on our built environments before they are recovered. Unfortunately, only a tiny portion
of these images are shared, curated, and utilized. There is a pressing need for a viable visual data organizing or categorizing tool with
a minimal manual effort. In this study, we aim to build a system to
automate classifying and analyzing a large volume of post-disaster
visual data. Our system called Automated Reconnaissance Image
Organizer (ARIO) is a web-based tool to automatically categorizing reconnaissance images using a deep convolutional neural net-
work and generate a summary report combined with useful metadata. Automated classifiers trained using our ground-truth visual
database classify images into various categories that are useful to
readily analyze and document reconnaissance images from post-
disaster buildings in the field.

Published
2020-01-02
How to Cite
Park, J., Yeum, C., Choi, J., & Liu, X. (2020). Automated Image Classification for Post-Earthquake Reconnaissance Images. Journal of Computational Vision and Imaging Systems, 5(1), 1. Retrieved from https://openjournals.uwaterloo.ca/index.php/vsl/article/view/1662