I have a task to extract buildings from satellite imagery of 0.5 m resolution and I don't have any elevation data to separate ground and elevated objects.. My question is how accurately we can extract building footprints with this available data using eCognition software? If anyone worked before this kind of data please let me know the process it helps me alot..

1 comment

  • Christian Weise

    I think you could reach an accuracy of 90% if you use CNN (in combination with refinement rules).

Recent Discussions

Need more help?

Enter Knowledge Base Contact Support