[Imports] NYC building and address import
remillard.jason at gmail.com
Wed Sep 11 03:49:56 UTC 2013
There are many way of doing this.... but if it was me, I would do this....
- load an OSM NY extract in a postGIS database with osm2pgsql
- for each building, run ST_Intersects with existing building=*, with
- if you don't get a hit, put the building into single directlyupload.osm
- if you do get a hit, use the current workflow, task manager, etc,
etc.. aka make a human deal with it...
On the directlyupload.osm file
- convert to osc -
- split to < 5,000 elements -
- batch upload -
Or coarse, you would need to do QA on the directlyupload.osm, looking
for schools without building tags, airport terminals without buildings
tags, and mostly for alignment. Once it is QA'ed, the file is
regenerated, and upload.
When the upload is done, Run this to
It will verify that you did not double upload anything by accident...
On Tue, Sep 10, 2013 at 10:05 PM, Ian Dees <ian.dees at gmail.com> wrote:
> Sorry, I misunderstood your comment. I'm happy to look at the tool you were
> talking about -- can you point us in the direction of where to find it
> and/or an example of its use?
> On Tue, Sep 10, 2013 at 9:00 PM, Jason Remillard <remillard.jason at gmail.com>
>> Hey, it is up to you! If you think it's not a big deal to spot check
>> million buildings, then go for it! I'm not helping either way.
>> And, yes, given who is on this team ... its 30 minute job for you
>> guys. For me, 3 hours.
>> On Tue, Sep 10, 2013 at 9:50 PM, Ian Dees <ian.dees at gmail.com> wrote:
>> > On Tue, Sep 10, 2013 at 6:12 PM, Jason Remillard
>> > <remillard.jason at gmail.com>
>> > wrote:
>> >> Hi,
>> >> NY city is so large (if it was a state, it would be 10th largest
>> >> state), any kind of manual work is going to be an big issue. Why not
>> >> just automated everything that is easy (say things that don't
>> >> conflict), or more aggressively don't merge correctly with Paul's
>> >> script, and let the humans focus on the hard addresses that can't be
>> >> easily sorted out by the automation. It would take you guys less than
>> >> a 30 minutes of coding to separate out the buildings that overlap with
>> >> OSM and those that don't. Push the buildings that don't overlap right
>> >> into OSM.
>> > Having done this in Chicago, I can say that it's not that difficult to
>> > do
>> > the manual conflation process. Three things:
>> > 1) There aren't that many buildings in OSM already. Probably more than
>> > there
>> > were in Chicago, but the majority of the tasks in task manager will
>> > simply
>> > be "spot check and upload"
>> > 2) In most cases, the buildings in OSM are simply tagged "building=yes"
>> > or
>> > similar. In these cases (unless there is noticeably better geometry in
>> > OSM),
>> > it's better to simply delete the data from OSM and let the imported data
>> > replace it because the imported buildings will probably have better/more
>> > geometry and tagging.
>> > 3) The process as they have it is pretty simple and easy to follow.
>> > Plus,
>> > the task manager is already loaded and ready to go. If it takes 30
>> > minutes
>> > of coding to separate out the buildings, I'm sure it would be helpful
>> > for
>> > you to go ahead and do that then make a pull request against the process
>> > in
>> > GitHub so discussion can happen. Don't forget that once this "upload
>> > totally
>> > new buildings" step is done then the task manager tasks would need to be
>> > removed or marked done, so don't forget to factor that into the time
>> > estimate for the automatic step.
>> > _______________________________________________
>> > Imports mailing list
>> > Imports at openstreetmap.org
>> > https://lists.openstreetmap.org/listinfo/imports
More information about the Imports