Land cover in Essen¶
Image source: Google Map Data (2021), GeoBasis-DE/BKG (2009)
Task¶
The goal of our first challenge is to classify a region south of Essen (Germany) into the following four land cover types:
Agriculture [1]
Forestry [2]
Urban [3]
Water [4]
To this end, you can use two Sentinel-2 scenes as predictor data, one from a winter overpass and one from a summer overpass. Additionally, we provide label data for a subset of the region covered that you should use to train your models. No further data sets are allowed in this challenge.
Label data was generated on the basis of the “Basis-DLM” from NRW’s opengeodata server.
Submission¶
You should hand in the prediction results for all pixels of the complete study area (733x1786 px) as one GeoTif-file. The classes must be encoded as integers of values 1 (agriculture), 2 (forestry), 3 (urban) and 4 (water). You can submit your results a maximum of 3 times. Only the best submission will be used for your ranking score.
Results must be submitted via Ilias. Please also stick to the following file name format:
YYYY-mm-dd_version_alias.tif
so for instance:
2021-12-01_V1_Catwoman.tif
Rating¶
The rating will be based on the accuracy score:
It will only be applied to a subset of the scene that is not part of the training data we provided.
