Ian Wulff-Limongi, Jaime Carrasco, Cristobal Pais, Alejandro Miranda, Andres Weintraub, Carla Vairetti and Diego Terán
Project on automatic recognition of fire scars using LANDSAT's satellite imagery applying the U-Net model
Wildfires are a critical problem among the last years worldwide due to their consequences, such as the carbon footprint, besides other ecological, economic and social impacts. Correspondingly, studying the affected areas is necessary, mapping every fire scar, typically using satellite data. In this paper, we propose a Deep Learning (DL) automate approach, using the U-Net model and Landsat imagery, that could become a straightforward automate alternative. Thus, two models were evaluated, each trained with a dataset with a different class balance, produced by cropping the input images to different sizes, to a determined and variable size: 128 and AllSizes (AS), including a better and worse class balance respectively. The testing results using 195 represen- tative images of the study area: Dice Coefficient (DC)=0.93, Omission error (OE)=0.086 and Commission Error (CE)=0.045- for AS, and DC=0.86, OE=0,12 and CE=0,12 for 128, proving that a better balanced dataset results on a better performance.
Two specific datasets were cropped out from the files of The Landscape Fire Scars Database, to evaluate the performance using different image sizes. These datasets included 1966 fires, dividing the data almost equally for each region, with 977 events from Valparaíso and 989 from BioBío.
Within the Convolutional Neural Network (CNN), the model U Net was selected for the prediction of the burned areas.
In the Table 1 can be seen the results for each model, AS and 128.
Finallly, some highlights of the models' performance can be seen:
To use the plugin in QGIS, follow the steps below:
Clone this repository into the QGIS plugins folder. On most systems, the folder is located at:
C:\Users\<username>\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins
Alternatively, you can clone the repository into any folder and create a symbolic link to the QGIS plugins folder.
Using the OSGeo4W Shell program, you need to create the resources.py file required by the plugin. Follow these steps:
-
Navigate to the plugin folder location:
cd <path_to_plugin_folder>
-
Ensure the correct virtual environments are activated by running the following commands:
- py3_env
- qt5_env
-
Generate the resources.py file from the resources.qrc file:
- pyrcc5 resources.qrc -o resources.py
- Restart QGIS if it was already open.
- Open the Plugins menu and navigate to the Manage and Install Plugins option.
- Find the plugin "Fire Scar Mapper" in the list and enable it.
Once the plugin is enabled, follow the interface prompts to:
- Select pre- and post-fire images. (If the images are not cropped) Provide a shapefile with fire scar boundaries or ignition points.
- Choose the model scale (AS or 128).
- Indicate whether the input images are already cropped.
- Run the plugin to generate fire scars directly within QGIS.
The plugin will generate georeferenced raster layers and organize them into groups for analysis.(The generated raster files will be stored on "/results" inside the plugin folder)