ESA-SatCen article on the use of Artificial Intelligence on SAR imagery

In the framework of the ESA-SatCen cooperation, a study has been carried out by the ESA Φ–Lab and the SatCen RTDI Unit (Research, Technology Development and Innovation), with the title "Deep Learning with Open Data for Desert Road Mapping". 

The work aims to demonstrate a methodology for road detection and monitoring in desert regions, using free input and reference data that can be scaled to desert regions globally. 

The approach takes input SAR data from open Copernicus Sentinel–1 satellite constellation over the area to be surveyed as well as available Open Street Map (OSM) data. An U–Net model has been trained with SAR amplitude and coherence averages, using OSM reference masks, for each desert region. The model was then applied to detect roads in each of the desert areas for which it was trained.

An article on the subject has been published in Remote Sensing and is available here: https://www.mdpi.com/2072-4292/12/14/2274

 

  • beach slider
    Image1 of 2Image of the SAR data analysis scenario: on the left side the processed SAR image used as input to the model, on the right side the roads detected by the model
  • beach slider
    Image2 of 2Roads detected by the model in a subset of the North Sinai AOI