Benchmark Dataset for Automatic Damaged Building Detection from Post-Hurricane Remotely Sensed Imagery

Citation Author(s):
Youngjun
Choe
University of Washington
Valentina
Staneva
University of Washington
Tessa
Schneider
Hertie School of Governance
Andrew
Escay
University of the Philippines
Christopher
Haberland
University of Washington
Sean
Chen
New York University
Submitted by:
Youngjun Choe
Last updated:
Wed, 09/01/2021 - 18:32
DOI:
10.21227/1s3n-f891
Data Format:
License:
Creative Commons Attribution
0
0 ratings - Please login to submit your rating.

Abstract 

Emergency  managers  of  today  grapple  with  post-hurricane damage assessment that is often labor-intensive, slow,costly,   and   error-prone.   As   an   important   first   step   towards addressing  the   challenge,   this   paper   presents   the   development of  benchmark  datasets  to  enable  the  automatic  detection  ofdamaged buildings from post-hurricane remote sensing imagerytaken  from  both  airborne  and  satellite  sensors.  Our  work  has two  major  contributions:  (1)  we  propose  a  scalable  framework to  create  benchmark  datasets  of  hurricane-damaged  buildings and  (2)  we  share  publicly  the  resulting  benchmark  datasets for  Greater  Houston  area  after  Hurricane  Harvey,  2017.  Thebenchmark  datasets  can  be  used  by  other  researchers  to  train and test object detection models which aim to detect the locationof  damaged  buildings  in  the  vast  imagery  over  affected  areas.

Instructions: 

Data can be used for object detection algorithms to properly annotate post disaster buildings as either damaged or non-damaged, aiding disaster response. This dataset contains ESRI Shapefiles of bounding boxes of buildings labeled as either damaged or non-damaged. Those labeled as damaged also have four degrees of damage from minor to catastrophic. Importantly, each bounding box is also indexed to one of the images in the NOAA post-Hurricane Harvey imagery dataset allowing users to match the bounding boxes with the correct imagery for training the algorithm. 

To make the NOAA imagery more manageable, images were processed and tiled into smaller 2048x2048 pixel ones. To obtain the same images please follow the steps below:

  1. Download the images from the NOAA page

  2. Tile the images using the tileTiff.py script (make sure the size is set to 2048 x 2048). All tiles will be in a subdirectory named “1”.

  3. This then creates the tiles that correspond to the image indexed in the shape files.

Important note: Not all bounding boxes in the shape file will map to an image. One will have to filter out bounding boxes which do not map to the images before feeding the data into any model.

Comments

for scientific research

Submitted by hao chen on Wed, 11/04/2020 - 22:34

for scientific research

Submitted by Muhammed Karademir on Wed, 11/11/2020 - 09:29

how to download the benchmark dataset?

Submitted by Swapandeep Kaur on Fri, 11/13/2020 - 12:32