Fig. 12: Original image subset (100 m × 100 m) and its augmented versions.
From: AI perceives like a local: predicting citizen deprivation perception using satellite imagery
Only moderate augmentations were used (vertical flip, 20% random rotation and 5% random translation, with missing values filled using the reflection mode) so that the augmented versions do not differ greatly from the original version. Satellite image subsets: WorldView-3 © 2019 Maxar Technologies.
