Damage-Map Estimation Using UAV Images and Deep Learning Algorithms for Disaster Management System

Introduction

The repo contains all source-code for our proposed approach in the paper entitled "Damage-Map Estimation Using UAV Images and Deep Learning Algorithms for Disaster Management System"

Forest fire information

Location: Andong, Republic of Korea, in April 2020

Date: from April 24, 2020 to April 26, 2020

Data acquisition information

Captured location: Mount 112, Ingeum-ri, Pungcheon-myeon, Andong-si, Gyeongsangbuk-do, 15-3 Haari, Namhu-myeon, Andong-si, Gyeongsangbuk-do

Devices: Phantom 4 Pro V2.0

Date: May 6, 2020

Burnt area mapping results

Setup environment

Install minianaconda

conda install -r requirements.txt

Sample data structure

The sample dataset shows how the implementation is carried out.

sample_data/sample_location_1_data
 |
 +-- Img
 |  |
 |  +-- img (1).png
 |  +-- img (2).png
 |  +-- ...
 +-- Label
 |  |
 |  +-- label (1).png
 |  +-- label (2).png
 |  +-- ...
 +-- Orig
 |  |
 |  +-- orig.JPG

sample_data/sample_location_2_data
 |
 +-- Img
 |  |
 |  +-- img (1).png
 |  +-- img (2).png
 |  +-- ...
 +-- Label
 |  |
 |  +-- label (1).png
 |  +-- label (2).png
 |  +-- ...
 +-- Orig
 |  |
 |  +-- orig.JPG

Scripts explaination

Use train_models to train the dual models.

With pretrained weights from this link , use predict_dual_models to predict sample testing dataset.

After receiving the predicted results, post processing functions can be used for postprocessing.

The EXIF information can be extracted by using extract EXIF function

Citation

If you use this code for your research, please cite our papers

@article{tran_damage-map_2020, title = {Damage-{Map} {Estimation} {Using} {UAV} {Images} and {Deep} {Learning} {Algorithms} for {Disaster} {Management} {System}}, volume = {12}, issn = {2072-4292}, url = {https://www.mdpi.com/2072-4292/12/24/4169}, doi = {10.3390/rs12244169}, number = {24}, journal = {Remote Sensing}, author = {Tran, Dai Quoc and Park, Minsoo and Jung, Daekyo and Park, Seunghee}, year = {2020} }