Optical and SAR change detection tools provide generic change detection maps for each pair of time-consecutive Sentinel-2 and respectively Sentinel-1 images that represent the same field of view. More concretely, these resulting maps are geoTiff rasters with the same size and geo-reference as the images on which they have been calculated, and whose pixel values range between 0 and 1 in order to represent the probability that a change has occurred between the acquisition dates of the two images.  

​Both tools are implemented on the CANDELA platform and accessible through Jupyter-notebooks documents available in the public folder.

Optical change detection tool

For the optical change detection tool, the notebook is divided in three parts:

  • A first part based on CreoDIAS library to access the sentinel-2 data of interest
  • A second part to run the change detection pipeline, which corresponds to the red path on the figure 
  • The last part that consists in training new neural network models and corresponds to the orange path on the image 1.

SAR change detection

For the SAR change detection tool, the notebook is divided in three parts:

  • A first part based on CreoDIAS library to access the sentinel-1 data of interest
  • A second part to preprocess the sentinel-1 data and crop them according to a geographical extent, which corresponds to the first chain on the first image
  • The last part to generate the change detection map, which corresponds to the second chain on the second image.

Please have a look at the D2.4 Deep Learning document and change detection videos if you want more information.

Thanks to the integration of a scheduler to have parallel computing, our change detection tool have allowed us to process entire France in only a few hours. These detections can benefit many use cases in order to save time, effort, and budget, such as urban or vineyard monitoring.