As @schem mentioned, your problem scenario falls under instance/semantic segmentation. In this case, your data is simply the satellite imagery you already have and the corresponding labels are the polygons which describe the instances (a.k.a pixels) which needs to be predicted for future satellite imagery that the model was not exposed before.

As @scherm mentioned, your problem falls under instance/semantic segmentation. In this case, your data is simply the satellite imagery you already have and the corresponding labels are the polygons which describe the instances (a.k.a pixels) which needs to be predicted for future satellite imagery that the model was not exposed to before.