Marc, Thanks for the not-so-positive feedback.
I'm a bit disappointed by the quality of wegenregister as a comparison basis (but I should have known better by now) but I'm still conviced about the general idea. In the beginning we had another subset of the data with only what we are relatively sure of coming from wegenregister but we noticed that the classifications don't make sense all the time. In the end we took all roads that had a name; at that point we discovered there are a lot of driveways in there with a name (ouwch!). It all comes down to: trying to detect missing stuff in OSM and I think there are some very good examples in the data for example these: http://maproulette.org/map/2789/3111729? http://maproulette.org/map/2789/3108600? We also ask **not to map when not sure** so if you look at the instruction you will see we don't ask to copy the data but to map whatever is missing, by the same standards as usual. If that's not possible remotely then don't map. We also shouldn't copy the streetnames except when we are sure it's an actual road. And yes sometimes that means a guestimate of what's a driveway, but how's that different from what we've been doing all along based on imagery? If anyone else has questions, feel free! Also, don't map when you're not sure! It's better not to have data, compared to having bad useless data. Cheers, Ben On Tue, Nov 14, 2017 at 9:37 AM, Marc Gemis <marc.ge...@gmail.com> wrote: > I've tried a couple of "tasks". > I found it pretty hard to fill in the proper road type without being > familiar with the local situation. > The fact that Wegenregister adds a name to each and every long drive > way is not going to help with the OSM data quality. > > So I gave up and went back to processing my surveys. > > m. > > On Tue, Nov 14, 2017 at 9:29 AM, joost schouppe > <joost.schou...@gmail.com> wrote: > > Hi, > > > > Ben and I have been working on "the road completion" project for quite > some > > time now. The idea is to build a toolchain that harvests any open data > > source on road networks, and turn it into a list of "possibly missing > roads > > in OSM". > > > > For a project overview, look here: > > http://www.osm.be/2017/01/06/en-project-road-completion.html > > > > Because it should be scalable and easy to adapt, we started with very > simple > > analysis methods. The idea is to make something simple that is quick to > set > > up and easy to extend. The repository is on Github [1] > > > > > > We started of with Wegenregister, an extensive road dataset for > Flanders. A > > first selection uses only roads that should be easy to map with just GRB > and > > aerial imagery. Because feedback is essential (to correct the many > mistakes > > in the source data, to fine-tune our processes, to avoid having a road > seem > > missing until someone uncritically adds it), we are using Maproulette to > > coordinate task completion. Instructions are available on the wiki [2] > > > > > > We're presenting this project at the Informatie Vlaanderen Trefdag. It's > a > > big thing that we got ourselves invited there at all! [3] > > > > If you want to help, start the task, have a look at the Github, improve > the > > documentation, find another dataset to use, etc! > > > > 1: https://github.com/osmbe/road-completion > > 2: > > https://wiki.openstreetmap.org/wiki/WikiProject_Belgium/ > Road_completion_project/Instructions > > 3: https://overheid.vlaanderen.be/trefdag-informatie-vlaanderen > > > > > > > > -- > > Joost Schouppe > > > > _______________________________________________ > > Talk-be mailing list > > Talk-be@openstreetmap.org > > https://lists.openstreetmap.org/listinfo/talk-be > > > > _______________________________________________ > Talk-be mailing list > Talk-be@openstreetmap.org > https://lists.openstreetmap.org/listinfo/talk-be >
_______________________________________________ Talk-be mailing list Talk-be@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-be
