[Imports] Importing Beirut building data

Tue May 2 09:47:40 UTC 2017

Hi folks - thanks for all of the replies.

I have a meeting with the AUB-NI folks to see what we can do about
licensing. From the sound of it, I don't think that they're strictly
against it.

Tracing over the entire city from aerial imagery is doable, but I'm afraid
about a few things:
1. The aerial imagery from Bing is mostly unusable, partly because it's
cloudy. Mapbox imagery is better, but not of a very high resolution.
2. The architecture in most of Beirut is not as clean and well-defined as
most European/North American cities. Buildings are separate but are *very*
tightly packed. The low resolution of the aerial images does not help with
this.
3. The building tools plugin for JOSM only draws rectangular buildings,
which is not the shape of most buildings in Beirut. Drawing the irregular
shapes of the buildings is very important, especially because there is no
commonly used address system here, so buildings (and their irregular
shapes) become used as landmarks for people to find places.

I'll let you know what is the result of my meeting with AUB-NI is.

Cheers,

/majd

On 2 May 2017 at 11:51, Rory McCann <rory at technomancy.org> wrote:

> Hi,
>
> So here's a suggestion, rather than doing a semi-manual "import" of this
> data, why not just trace the buildings manually from aerial imagery?
> You're suggesting a very manual "import" process anyway, so I don't
> think a proper tracing mapping party would take much longer? You can use
> something like the Tasking Manager, or MapCraft to split the city up
> into "cake slices" and just go through them one by one, mapping buildings.
>
> The JOSM Building Tools plugin is fantastic and can allow you to map
> buildings very quickly.
>
> This has the advantage of not being an import, and you don't need to
> think about licencing, you don't need to write software to "show a
> random building". You'd be surprised at how fast some people can map
> once they get the hang of it.
>
> Rory
>
> On 28/04/17 15:35, Majd Al-shihabi wrote:
>
>> Hi folks,
>>
>> I have been working with an organisation called Public Works Studio to
>> create a map of housing evictions in Beirut over the past couple of
>> years. As a part of it, we need a good base map containing buildings in
>> the city, but unfortunately, OSM doesn't have that yet.
>>
>> Through our network, we found out that a group at the American
>> University of Beirut called The Neighbourhood Initiative (AUB-NI) has
>> done a survey of all of the buildings in the city. The data can be
>> viewed here
>> <https://scholarworks.aub.edu.lb/handle/10938/10283?show=full>. As far
>> as I understand (and of course i'm not a legal expert), the license on
>> the data from the AUB-NI is not incompatible with the Open Database
>> License. In any case, my understanding is that they got the data freely
>> from the municipality.
>>
>> Unfortunately, the Lebanon OSM mailing list has been inactive since
>> 2012, but at a recent HOT training, and through my work, I have met a
>> few people who are interested in activating the community again.
>>
>> What we would like to do is to start importing all 20k+ building
>> polygons into OSM, but do that in a special way. We would like to run a
>> series of mapathons where people enter the buildings one by one. We
>> would create a simple tool with the following workflow:
>>
>>  1. Display to the mapper a random building from the available 20k+
>>     buildings
>>  2. Overlay the building on the mapbox satellite images (Mapbox is
>>     clearer than Bing)
>>  3. If
>>       * the building already exists in OSM, discard the new polygon
>>         (maybe we can avoid this option by doing some cleaning of the
>>         data using QGIS)
>>       * the building does not exist on OSM, and the polygon matches the
>>         satellite image, then add as many tags as we know, and upload it
>>         to OSM
>>       * the building does not exist on OSM, and the polygon does NOT
>>         match the satellite image, then modify the polygon so it
>>         matches, add as many tags as possible, then upload to OSM
>>
>> This process serves an important function: to verify the accuracy of the
>> data from the AUB-NI, especially since the dataset is from 2004, and the
>> city has changed A LOT since that date.
>>
>> From a technical point of view, I can see three options to do this:
>> 1. The simplest way is to find a way to pass the polygons to the iD
>> editor (through GET or POST parameters) and have it preselected for the
>> user to make any modifications to it.
>> 2. Alternatively, we could build an interface that would save the
>> polygon to OSM *and then* ask the user to verify it.
>> 3. Otherwise (and least desirable option) is to build an entire
>> javascript app that would implement the workflow.
>>
>> I've done a bit of research and option 2 seems like it's the most
>> feasible one, but I'd like to hear some more thoughts from the community
>> about this.
>>
>> Many thanks for your advice, in advance :)
>>
>> /majd
>>
>>
>> _______________________________________________
>> Imports mailing list
>> Imports at openstreetmap.org
>> https://lists.openstreetmap.org/listinfo/imports
>>
>>
>
>
> _______________________________________________
> 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/20170502/73b1b64f/attachment.html>