Well, you have to consider 2 contexts: - OSM data + stylesheet used in training phase which would not really be a derivative work IMO - no way model could memorize OSM data and the stylesheet is CC0 anyway - OSM data as fed to the network during inference phase
I think whatever the status is, it would be comparable to geocoding, yet bearing even less resemblance to source data. In some way, looking at the output as a whole, it's a 16 channel very low resolution render. On Mon, Jun 18, 2018 at 6:25 PM, Mateusz Konieczny <[email protected]> wrote: > 18. Jun 2018 10:39 by [email protected]: > > One interesting thing about this kind of work is the legal side and the > question if a process as described there is a derivative work of the > training data used. > > > I am surprised that somebody may think that it is not a derivative work > of training dataset. > > _______________________________________________ > dev mailing list > [email protected] > https://lists.openstreetmap.org/listinfo/dev > >
_______________________________________________ dev mailing list [email protected] https://lists.openstreetmap.org/listinfo/dev
