Mapping buildings

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Mapping buildings

john whelan-2
We tend to think of mapping buildings as one of the simpler tasks we ask mappers to do but recently I've been looking at code that can extract buildings from OpenStreetMap aggregate the floor area then calculate an estimated population.

Trouble is when I look at the map in some areas the buildings are accurately​ mapped but in others the standard of mapping leaves much to be desired.  Validation is a problem.  Do we expect validators to carefully move the four points to the correct corners and square them?  This is more effort than remapping them with the JOSM building_tool plugin and is unlikely to happen.  I also come a cross a large number of settlements tagged building=yes or building=residential these I correct as I come across them.

Round huts are a particular problem, it's very easy to copy one hut which means they often end up being mapped the same standard size.

Perhaps if we explain why it is important to map buildings accurately in the instructions we might get better results.

Dunno, perhaps answer is not to estimate population based on floor area.

Thoughts?

Thanks John 

_______________________________________________
HOT mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/hot
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Mapping buildings

Daniel O'Connor
I tend to use osmose to detect 'large building intersection clusters' at a very large scale - IE during the Nepal activation.
Others that show up building-in-building or 'special building due to size marked building=yes' are also useful QA checks, often catching whole residential areas marked as a building for example.

I think those may be the low hanging fruit of validation; and worth considering as a recommended check at the project level.
Would be interesting to integrate those results into a 'validations' tab per tile or per project, keyed off of an osmose feed.

Beyond that, its reasonably common to encourage 'squaring' the buildings - but I dont think you'd be able to push much further than that and get the majority doing it by default (would have to make it easier to one click improve the traced data somehow)

On 20 Mar 2017 7:47 AM, "john whelan" <[hidden email]> wrote:
We tend to think of mapping buildings as one of the simpler tasks we ask mappers to do but recently I've been looking at code that can extract buildings from OpenStreetMap aggregate the floor area then calculate an estimated population.

Trouble is when I look at the map in some areas the buildings are accurately​ mapped but in others the standard of mapping leaves much to be desired.  Validation is a problem.  Do we expect validators to carefully move the four points to the correct corners and square them?  This is more effort than remapping them with the JOSM building_tool plugin and is unlikely to happen.  I also come a cross a large number of settlements tagged building=yes or building=residential these I correct as I come across them.

Round huts are a particular problem, it's very easy to copy one hut which means they often end up being mapped the same standard size.

Perhaps if we explain why it is important to map buildings accurately in the instructions we might get better results.

Dunno, perhaps answer is not to estimate population based on floor area.

Thoughts?

Thanks John 

_______________________________________________
HOT mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/hot



_______________________________________________
HOT mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/hot
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Mapping buildings

Jo-2
In JOSM you can use the 'Improve way accuracy' tool (w) to conveniently drag those corners followed by 'q' to square the building. That is somewhat less time consuming than redrawing (and it keeps the history). A validator can choose to either do this, or to invalidate the task. It's probably a good idea to include a link to a video explaining how to fix the problem, in case the square gets invalidated.

Concerning the round buildings, it would be good to make mappers aware of the possibility to use Ctrl-Alt-drag left mouse button to rescale after using Ctrl-d. OK, that's also a JOSM thing. I think we should instruct people to use JOSM for HOT mapping in the first place or finally implement this kind of useful functionality in iD. Maybe there is still time to propose it for GSoC? I won't propose it, as I'll only propose things that I can mentor myself.

Polyglot

2017-03-20 1:31 GMT+01:00 Daniel O'Connor <[hidden email]>:
I tend to use osmose to detect 'large building intersection clusters' at a very large scale - IE during the Nepal activation.
Others that show up building-in-building or 'special building due to size marked building=yes' are also useful QA checks, often catching whole residential areas marked as a building for example.

I think those may be the low hanging fruit of validation; and worth considering as a recommended check at the project level.
Would be interesting to integrate those results into a 'validations' tab per tile or per project, keyed off of an osmose feed.

Beyond that, its reasonably common to encourage 'squaring' the buildings - but I dont think you'd be able to push much further than that and get the majority doing it by default (would have to make it easier to one click improve the traced data somehow)

On 20 Mar 2017 7:47 AM, "john whelan" <[hidden email]> wrote:
We tend to think of mapping buildings as one of the simpler tasks we ask mappers to do but recently I've been looking at code that can extract buildings from OpenStreetMap aggregate the floor area then calculate an estimated population.

Trouble is when I look at the map in some areas the buildings are accurately​ mapped but in others the standard of mapping leaves much to be desired.  Validation is a problem.  Do we expect validators to carefully move the four points to the correct corners and square them?  This is more effort than remapping them with the JOSM building_tool plugin and is unlikely to happen.  I also come a cross a large number of settlements tagged building=yes or building=residential these I correct as I come across them.

Round huts are a particular problem, it's very easy to copy one hut which means they often end up being mapped the same standard size.

Perhaps if we explain why it is important to map buildings accurately in the instructions we might get better results.

Dunno, perhaps answer is not to estimate population based on floor area.

Thoughts?

Thanks John 

_______________________________________________
HOT mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/hot



_______________________________________________
HOT mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/hot



_______________________________________________
HOT mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/hot
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Mapping buildings

Kretzer

I didn't know about the JOSM shortcut. Thanks, that's  very useful!

Also Iove the idea of having a scaling tool in iD - would second that suggestion.
On the other hand I don't have much practical use for the new mirroring tools. Maybe others have the same priorities?



Am 20.03.17 um 08:29 schrieb Jo

> In JOSM you can use the 'Improve way accuracy' tool (w) to conveniently
> drag those corners followed by 'q' to square the building. That is somewhat
> less time consuming than redrawing (and it keeps the history). A validator
> can choose to either do this, or to invalidate the task. It's probably a
> good idea to include a link to a video explaining how to fix the problem,
> in case the square gets invalidated.
>
> Concerning the round buildings, it would be good to make mappers aware of
> the possibility to use Ctrl-Alt-drag left mouse button to rescale after
> using Ctrl-d. OK, that's also a JOSM thing. I think we should instruct
> people to use JOSM for HOT mapping in the first place or finally implement
> this kind of useful functionality in iD. Maybe there is still time to
> propose it for GSoC? I won't propose it, as I'll only propose things that I
> can mentor myself.
>
> Polyglot
>
> 2017-03-20 1:31 GMT+01:00 Daniel O'Connor <[hidden email]>:
>
> > I tend to use osmose to detect 'large building intersection clusters' at a
> > very large scale - IE during the Nepal activation.
> > Others that show up building-in-building or 'special building due to size
> > marked building=yes' are also useful QA checks, often catching whole
> > residential areas marked as a building for example.
> >
> > I think those may be the low hanging fruit of validation; and worth
> > considering as a recommended check at the project level.
> > Would be interesting to integrate those results into a 'validations' tab
> > per tile or per project, keyed off of an osmose feed.
> >
> > Beyond that, its reasonably common to encourage 'squaring' the buildings -
> > but I dont think you'd be able to push much further than that and get the
> > majority doing it by default (would have to make it easier to one click
> > improve the traced data somehow)
> >
> > On 20 Mar 2017 7:47 AM, "john whelan" <[hidden email]> wrote:
> >
> > We tend to think of mapping buildings as one of the simpler tasks we ask
> > mappers to do but recently I've been looking at code that can extract
> > buildings from OpenStreetMap aggregate the floor area then calculate an
> > estimated population.
> >
> > Trouble is when I look at the map in some areas the buildings are
> > accurately​ mapped but in others the standard of mapping leaves much to be
> > desired.  Validation is a problem.  Do we expect validators to carefully
> > move the four points to the correct corners and square them?  This is more
> > effort than remapping them with the JOSM building_tool plugin and is
> > unlikely to happen.  I also come a cross a large number of settlements
> > tagged building=yes or building=residential these I correct as I come
> > across them.
> >
> > Round huts are a particular problem, it's very easy to copy one hut which
> > means they often end up being mapped the same standard size.
> >
> > Perhaps if we explain why it is important to map buildings accurately in
> > the instructions we might get better results.
> >
> > Dunno, perhaps answer is not to estimate population based on floor area.
> >
> > Thoughts?
> >
> > Thanks John
> >
> > _______________________________________________
> > HOT mailing list
> > [hidden email]
> > https://lists.openstreetmap.org/listinfo/hot
> >
> >
> >
> > _______________________________________________
> > HOT mailing list
> > [hidden email]
> > https://lists.openstreetmap.org/listinfo/hot
> >
> >
> _______________________________________________
> HOT mailing list
> [hidden email]
> https://lists.openstreetmap.org/listinfo/hot

_______________________________________________
HOT mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/hot
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Mapping buildings

Jo-2
I created a document to help people coming to our National Mapathon containing a fast introduction to JOSM:


2017-03-21 22:37 GMT+01:00 Kretzer <[hidden email]>:

I didn't know about the JOSM shortcut. Thanks, that's  very useful!

Also Iove the idea of having a scaling tool in iD - would second that suggestion.
On the other hand I don't have much practical use for the new mirroring tools. Maybe others have the same priorities?



Am 20.03.17 um 08:29 schrieb Jo

> In JOSM you can use the 'Improve way accuracy' tool (w) to conveniently
> drag those corners followed by 'q' to square the building. That is somewhat
> less time consuming than redrawing (and it keeps the history). A validator
> can choose to either do this, or to invalidate the task. It's probably a
> good idea to include a link to a video explaining how to fix the problem,
> in case the square gets invalidated.
>
> Concerning the round buildings, it would be good to make mappers aware of
> the possibility to use Ctrl-Alt-drag left mouse button to rescale after
> using Ctrl-d. OK, that's also a JOSM thing. I think we should instruct
> people to use JOSM for HOT mapping in the first place or finally implement
> this kind of useful functionality in iD. Maybe there is still time to
> propose it for GSoC? I won't propose it, as I'll only propose things that I
> can mentor myself.
>
> Polyglot
>
> 2017-03-20 1:31 GMT+01:00 Daniel O'Connor <[hidden email]>:
>
> > I tend to use osmose to detect 'large building intersection clusters' at a
> > very large scale - IE during the Nepal activation.
> > Others that show up building-in-building or 'special building due to size
> > marked building=yes' are also useful QA checks, often catching whole
> > residential areas marked as a building for example.
> >
> > I think those may be the low hanging fruit of validation; and worth
> > considering as a recommended check at the project level.
> > Would be interesting to integrate those results into a 'validations' tab
> > per tile or per project, keyed off of an osmose feed.
> >
> > Beyond that, its reasonably common to encourage 'squaring' the buildings -
> > but I dont think you'd be able to push much further than that and get the
> > majority doing it by default (would have to make it easier to one click
> > improve the traced data somehow)
> >
> > On 20 Mar 2017 7:47 AM, "john whelan" <[hidden email]> wrote:
> >
> > We tend to think of mapping buildings as one of the simpler tasks we ask
> > mappers to do but recently I've been looking at code that can extract
> > buildings from OpenStreetMap aggregate the floor area then calculate an
> > estimated population.
> >
> > Trouble is when I look at the map in some areas the buildings are
> > accurately​ mapped but in others the standard of mapping leaves much to be
> > desired.  Validation is a problem.  Do we expect validators to carefully
> > move the four points to the correct corners and square them?  This is more
> > effort than remapping them with the JOSM building_tool plugin and is
> > unlikely to happen.  I also come a cross a large number of settlements
> > tagged building=yes or building=residential these I correct as I come
> > across them.
> >
> > Round huts are a particular problem, it's very easy to copy one hut which
> > means they often end up being mapped the same standard size.
> >
> > Perhaps if we explain why it is important to map buildings accurately in
> > the instructions we might get better results.
> >
> > Dunno, perhaps answer is not to estimate population based on floor area.
> >
> > Thoughts?
> >
> > Thanks John
> >
> > _______________________________________________
> > HOT mailing list
> > [hidden email]
> > https://lists.openstreetmap.org/listinfo/hot
> >
> >
> >
> > _______________________________________________
> > HOT mailing list
> > [hidden email]
> > https://lists.openstreetmap.org/listinfo/hot
> >
> >
> _______________________________________________
> HOT mailing list
> [hidden email]
> https://lists.openstreetmap.org/listinfo/hot


_______________________________________________
HOT mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/hot
Loading...