Skip to content

Latest commit

 

History

History
9 lines (5 loc) · 811 Bytes

README.md

File metadata and controls

9 lines (5 loc) · 811 Bytes

Deep Neural Network Regression for Digital Surface Model Generation with Sentinel-2 Imagery

This is the code of the paper "Deep Neural Network Regression for Digital Surface Model Generation with Sentinel-2 Imagery". The repository contains all three model variants (branches) of U-Net (V1-Baseline, V2-Multiscaled and V3-Attention) and our model for comparison, Adabins, in the within the four branches including the all scripts supporting the network regarding data preparation or postprocessing. Please note, that V3-Attention is refered with a v4 suffix in the files. The finished paper can be found here: https://ieeexplore.ieee.org/document/10189905

Kind regards,

Konstantin Müller, Robert Leppich, Christian Geiß, Vanessa Borst, Patrick Aravena Pelizari, Samuel Kounev and Hannes Taubenböck