ML/AI building footprints often flag shapes that aren't buildings at all. The footprint in question can be compared to nearby features in an OSM database to guess the probability the footprint is a buildings. Any feature identified as such get a unique tag added so they can be easily found using JOSM for manual validation.
ML/AI building footprints often flag shapes that aren't buildings at all. The footprint in question can be compared to nearby features in an OSM database to guess the probability the footprint is a buildings. Any feature identified as such get a unique tag added so they can be easily found using JOSM for manual validation.