"To produce these photo tours, we use advanced computer vision techniques to create a 3D experience from public, user-contributed Picasa and Panaromio photos. We start by finding clusters of overlapping photos around major landmarks. From the photos, our system derives the 3D shape of each landmark and computes the location and orientation of each photo. Google Maps then selects a path through the best images, and adds 3D transitions to seamlessly guide you from photo to photo as if you're literally flying around the landmark and viewing it from different perspectives," explains Google.
