Example of Service Usage

Description

This section describes how to produce a seabed classification map from Multispectral Multibeam Data. The models that the service provide have been trained using a Random Forest classifier considering three, four and five individual seabed classes, respectively. More details regarding the training e can be found in the relevant publication [1]. The service can be used either by providing your own data or by using demo data provided with the service [2].

Uploading your own Multispectral GeoTIFF file

To upload your own file (3-band Georeferenced TIFF file), select a file located on your local machine using the provided form and then clicking on the “Upload File” button.

_images/UI.png

Produce classification map

The procedure to produce the classification map is described below.

  1. Select the file you want to classify in the list of available files
  2. Select the classification model you want to use according to the desired number of seabed Classes.
  3. Press the “Classify Image” button at the bottom of the page.
  4. Inspect the results at the bottom of the page
  5. Click the “Download result” button if you wish to keep the results.

An example of the results is presented below using three classes model which contains the classes:

  • Sand
  • Mud Sand
  • Gravel

(Top Left: Multispectral Trained Image, Top Middle: Classification Map (3 Classes), Top Right: Bathymetry of the area and at the bottom three representative images of each class)

_images/results.png

Note

If you are facing any issue, please send us an email: panag.mertikas@gmail.com

[1]Mertikas, P., & Karantzalos, K. (2020). Seafloor Mapping from Multispectral Multibeam Acoustic Data at the European Open Science Cloud. The International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences, 43, 985-990.
[2]Demo data collected at Patricia Bay, Canada