I'm really looking forward to read the procedure on how you guys came up. You might have been lucky that I couldn't make it , now you only have to ignore my annoying mails :-)
I fail to see how documentation will prevent any participant from just dumping the data from GRB into OSM without any check what so ever. The same happened in The Netherlands where destroyed buildings were still imported and not-yet existing house numbers where dumped in the dunes. A model import does not introduce errors that could be avoided by simply looking at aerial imagery I've been working almost daily with the GRB data now for 6 months (*) or so and have a pretty good idea how much work it takes to make this a model import. I fail to see how 20-25 people can accomplish this is a year, unless they really spend a lot of hours on it. But I am probably to critical right now and have to see what process / documentation you guys came up with. so I shut up now and wait regards m (*) I did add new data to complement my surveys (super easy), as well as update existing data in my neighborhood and on many churches (sometimes very hard to do). Keeping the history and data of a building, keeping nodes attached to existing buildings (entrances or maiboxes), existing, but differently split-up of the buildings will all slow you down, because each time you have to think how you have to solve that problem. Not to mention landuses, amenities, etc that are attached to the buildings and have to be redrawn. Without doing all of this, no model import. On Wed, Dec 14, 2016 at 9:51 PM, Glenn Plas <gl...@byte-consult.be> wrote: > On 14-12-16 20:55, Marc Gemis wrote: >> On Wed, Dec 14, 2016 at 8:49 AM, joost schouppe >> <joost.schou...@gmail.com> wrote: >>> Expect between 10 and 20 users for the import tool. I think we can handle a >>> little ugliness. >>> So something "not too bad" will do :) >>> >>> That said, if it all goes as planned, we will be doing a "model import". So >>> I think the work on >> >> A model import of this size and data quality with only 10-20 users ? >> You must be kidding :-) > > Sure, I would to it alone in 6Months in madness mass-mode approach. In > Escada approach with eye for detail and survey information it would > take a lot but there is not due date planned. It's not an import, its a > merge tool. > >> Without involvement of really local mappers (1 per village), it is >> IMHO impossible to bring the GRB data up to level that is needed for >> a "model import". >> >> As you might know, everything with an address is categorized as house, >> everything without as shed. While this is a choice that Glenn made, >> there are only 2 categories you can start from. There are also roofs >> and some buildings are known as industrial (substations for >> electricity), but that does not help with the issue. > > It's not a choice by me, it's a source data driven decision , you can't > make up 5 categories when the source is only 2 big. But we actually > discussed what we will do with those cases and explained why it is what > it is. There are plenty of solutions for these cases but since it's a > merge tool you're discussion a well balanced/chosen default based on the > source data, we can/will modify that a bit, there are a few new ideas > now since sunday. But all in all, this is a default, you get the merge > windows which is basically what a mapper does: make informed decisions > with a certain amount of local knowledge. > > >> >> So even churches are marked as shed. While this type of correction is >> easy, it might easily be skipped (see Urbis import). However, try to >> recognize apartments, retail vs. commercial, warehouse vs. industrial, >> shed vs. garage (next to house), civic, barns vs. stables vs >> farm_auxilairy etc. buildings based on aerial imagery. > > Yes, again, we have discussed this and we have a solution for this, this > hackaton was exactly for bringing this up, you are a power user but you > missed the meeting unfortunately, it happens but now I feel like again > we need to discuss this, we are aware of the issues but we have brained > together solutions for this. > >> >> Unless many, many more mappers get involved in the process, this will >> not be a model import in my eyes, simply because too many buildings >> that are not house/shed will be mapped as such. > > 20 mappers you can do Belgium in a single year. Not every mapper is > going to work as detailed as you , but for GRB we need it to be done well. > > >> >> BTW, this has nothing to do with the import tool, which even in it's >> current version is good to support local mappers. The main benefit is >> that you do not have to draw the buildings yourself or add the >> addresses. That's already done, changing the building type based on >> ones observations is easy. >> >> So I really hope that the people involved in the import realize that >> improving the data before uploading is much more important that having >> an exact copy of GRB data in OSM. > > > Yea, I really missed you there Marc, you would have been an asset. You > have good input but missed the party this time. We know it's not easy > to make it sometimes but give us some time to document things and then > give your feedback again on those solutions. > > > _______________________________________________ > 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
