4 ms·

I still wonder at which point we can use machine learning to auto-extend OSM maps based on satellite images. Or even better, at which point Tesla on-the-fly dat

by faebi 4y ago

I still wonder at which point we can use machine learning to auto-extend OSM maps based on satellite images. Or even better, at which point Tesla on-the-fly data can be used to create persistent maps. I mean just have a look at the visible amount of data in that link. You literally can see the swans. https://www.openstreetmap.org/edit#map=22/47.36686/8.54401 https://www.openstreetmap.org/edit#map=22/47.36686/8.54401 Ignoring Tesla FSD itself, just look at the generated map data. Imagine how much more we can get if we don't need to run this at 20fps. https://www.youtube.com/watch?v=wQkXcySUnJk https://www.youtube.com/watch?v=wQkXcySUnJk

habi 4y ago
> I still wonder at which point we can use machine learning to auto-extend OSM maps based on satellite images. There’s https://wiki.openstreetmap.org/wiki/RapiD https://wiki.openstreetmap.org/wiki/RapiD which helps users to add features based on AI predictions. Each feature has to be properly reviewed though. A blank import of such features won’t happen anytime soon. > I mean just have a look at the visible amount of data in that link. You literally can see the swans. https://www.openstreetmap.org/edit#map=22/47.36686/8.54401 https://www.openstreetmap.org/edit#map=22/47.36686/8.54401 The swisstopo imagery is one of the better ones you can get, but swans should not be mapped in OSM ;)