I did pretty much the same thing some 10 years ago but only for my own images and this creator overcame a problem that I could not. I ingested a collection of images, read the metadata and used that to place the images on a map like this. However, I could not figure out how to determine facing. This project has an arrow indicator to show this, I ended up setting it manually. I would love to know how they determined this because I really cannot figure it. Does image metadata contain more information than it used to? Is there a way to use an AI agent to analyze the light/angles/streets/landmarks to approximate this?
As far as I know all images are manually uploaded and geotagged. Since posting this I've uploaded tens of photos of my town. I've had to research every photo, using current Street View, historical maps (Sanborn Fire Insurance maps are crucial), even going as far as looking up census records to find out where people were living in the case of house photos. It's been a ton of fun, and I've learned so much about my town, and generally how records are kept.
Ze Memoriez!
But why is it in Russian, even when EN is active?
Can be compensated somewhat by chosing a different OSM-layer, though.