Satelite images are the ground truth OSM is traced on. And the quality of those tracings varies wildly at times, I've had to fix weirdly offset shores that had roads on them placed on the sea on more than one occasion.
If this can be somewhat consistent then it'll probably do better than the average OSM contributor. Something like segmenting houses, roads, bodies of water, comparing against current data and highlighting inconsistencies for correction would be a good start though.
> Something like segmenting houses, roads, bodies of water, comparing against current data and highlighting inconsistencies for correction would be a good start though.
Hi there, I agree that is a valuable usage of a model trained with OSM data.
I didn't have the time to release the code but I am/was doing exactly that to refine the training dataset.
I take the trained model and run it against the ground truth from OSM. Any heavy mismatch between the two almost always result in an useful edit to be made in OSM.
> it'll probably do better than the average OSM contributor
let us reconsider this statement, please. An unexpected and powerful effect of the Openstreetmap project is iterated convergence on ground truth. No person is perfect in contribution, and few people are consistently terrible. Revision and updates, common vision towards accuracy, an appreciation of cooperative contributions.. have astounded the public and humbled critics repeatedly. Not because every key stroke and mouse click is perfect, but because iteration and plural sources have converged in a usable system of software and data.
AI inputs to Openstreetmap are not new, as noted in other comments. The path forward is bright, useful to humans and participatory in the Openstreetmap project.
I was on a search for the 'perfect' system all my life.
And I found it after 40 years : The No-System System... and it's the exactly opposite of op's suggestion :-)
All together in somewhat chaotic folders and subfolders... the clou is that I use "recoll" everytime i search something. It's an Index based search engine. Take a look...
I never missed something since I use just recoll and throw things just anywhere in the huge black box.
The main pro: it costs me no time to "sort" things into things.
At least in Germany, such agreements are afaik invalid and without a severability clause, possibly all others too. Simply because something like copyright cannot be assigned in Germany. Secondly, there are ways to use Reddit without ever having agreed to the ToS.
That clause does not assign copyright. You explicitly keep your own copyright (in the previous clause, I didn't reproduce it above). You just grant them a license to use your content in the ways they listed.
In Germany lopsided contracts clauses that surrender all rights are void. GDPR also gives you the right to recall all your data, so you should be able to delete your account and all your data.