OneSoil uses Sentinel-2 satellite images with a 10-meter resolution. Data on different regions of the world is updated every three to five days. All satellites photos are cleaned of snow, clouds, shadows and other artifacts; this ensures the accuracy of the subsequent operations. To develop an automatic field recognition technology, OneSoil manually marked out several tens of thousands of fields. Then the Data Science team built the machine learning algorithm that helped OneSoil to find
60 million fields in the US and Europe.