Investigation and validation of an objective method for analysis of soil coverage at tillage


The project targets to develop an optical method for objective estimation of soil cover. A machine learning technique is developed which segments a soil image into divert soil cover types. For easy mobile usage, the algorithm is integrated into a web service. The estimate soil cover value delivers a measurement for the corrosion resistance of the soil. For the validation of the system, manual annotated images are compared with its classification results from the developed algorithm.


Duration:  01/2015 – 05/2017
  • IKT Petzenkirchen
  • Farmdok GmbH Wieselburg
Funded by: Land Niederösterreich



Dipl.-Ing. Dr. Peter Riegler-Nurscher JR: Computer Vision & Machine Learning
Rottenhauser Straße 1
3250 Wieselburg