Searching an aerial photo with text queries - a demo and how it works
Briefly

Vector embeddings allow searching aerial photos without using map data, interpreting text queries like 'roundabout' or 'tennis court' to provide relevant image results.
SkyCLIP model, trained on image-text pairs, creates 768-element vectors that can be used to find similarities between different objects in aerial photos, enabling accurate searches.
Read at Rtwilson
[
|
]