[Imports] Tanzania buildings

Rob Nickerson rob.j.nickerson at gmail.com
Fri Apr 25 18:23:36 UTC 2014


Yes, when I looked at it in QGIS the other day I also found an offset. I'm
not great with projections so it might be solvable with ease or could be
tricky.

As for building shapes, the small section I looked at seemed OK. Not to the
standard I aim for when tracing, but better than other mappers tracing
standard (building tracing is a tedious task). Personally I think its up to
the local community as it's them who will be doing any clean up.

Definitely need to keep existing OSM buildings/tags.

Rob
On 25 Apr 2014 04:30, "Mikel Maron" <mikel_maron at yahoo.com> wrote:

> Hi Msilikale, Rob
>
> Msilikale, thanks for the initiative of sharing the Dar building footprint
> data set (which btw is posted at
> https://www.dropbox.com/sh/oe08t8tirw7qtuk/AiOhiUBEN5). It is certainly
> comprehensively covering the whole metro region, and beyond, massive!
> Actually, can you explain the boundary of data set? Seems like the urban
> expanse to the south is suddenly cut.
>
> I took a look at this data set, compared to what's presently in
> OpenStreetMap. Certainly much greater coverage in the government data set.
> But many of the buildings in OSM have other valuable tags. Any import
> process would want to retain these.
>
> Additionally, there appears to be a consistent offset in this data set.
> I'm not sure if it's an artifact of the reprojection, or in some
> misalignment in the georeferencing of the source imagery for this building
> data set (I assume it was digitized from imagery); or if there is
> misalingment in Bing imagery in Dar (though I suspect not, both Google and
> Bing align decently well).
>
> Looking at the geometries of the building footprints themselves, some look
> decently accurate relative to the imagery, but in a great majority of cases
> there is not correspondence, in a number of ways. Sometimes an individual
> building is broken up into several. Other times, the footprint is partially
> correct, but then in other parts wrong. And in other cases, there's clearly
> been new construction. This is mostly looking in the Central District. If I
> look at some place like Tandale, there seems to be even less correspondence
> (to be fair, OSM is not much better here).
>
> If this data were to be imported, it would need to be looked at very
> carefully, piece by piece. It couldn't be done in bulk I think. Could be a
> big detailed job. And I wonder, would it end up being more work to try to
> get this data set to fit the existing OSM data, and imagery, than simply
> organizing to trace buildings directly from imagery. Could even possibly be
> an interesting microtasking exercise, ala
> http://buildinginspector.nypl.org/
>
> In any case, Msilikale, worth reading through the Import guidelines to lay
> out some of the other considerations, and good next steps, like documenting
> this data set and deliberation in the wiki.
> http://wiki.openstreetmap.org/wiki/Import/Guidelines
>
> Best
> Mikel
>
> * Mikel Maron * +14152835207 @mikel s:mikelmaron
>   On Monday, April 21, 2014 5:45 PM, Rob Nickerson <
> rob.j.nickerson at gmail.com> wrote:
>
> Hi Msilikale,
>
> Thanks for uploading the Tanzania, Dar es salaam buildings data to
> dropbox. I've downloaded it and had a quick look. The tags look like they
> are not needed with the only one that *may* be useful being called "BIN" -
> I have no idea what it is. Do you know?
>
> I will split the file into smaller sections for you and give you a link to
> the download. Please give me a few days. I will aim to get it done by the
> end on next weekend. If not then feel free to contact me to remind me.
>
> Rob
>
>
>
> _______________________________________________
> Imports mailing list
> Imports at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/imports
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/imports/attachments/20140425/6523ee78/attachment.html>


More information about the Imports mailing list