How can explainable AI be used to understand what constitutes wilderness?

On July 8th Timo Stomberg, KI:STE doctoral researcher, presented jUngle-Net at the ISPRS Congress 2021. jUngle-Net is an explainable neural network which we use to gain new insights into the appearance of wilderness in satellite images.

Check out Timo’s explanation on Youtube jUngle-Net  and read the paper here: 

by Timo Stomberg