Automated Road Segmentation for Change Detection in Cologne

Slider comparisons of the models segmentations for the Cologne dataset's five locations

Refresh page if slider images dont appear on first load. Left images: 1998. Right images: 2019

         1                2                3                4                5

    Dice coefficient: 0.3109 - Loss: 0.6891- Recall: 0.3123 - Precision: 0.3842