I strongly agree that a locally sourced dataset with more tags would be preferred in osm. Adding just building outlines complicates future imports which might be higher quality. If you want help acquiring a better dataset, asking on this list or a list dedicated to the specific region would be a good start.
I'm generally opposed to adding source tags on elements. I believe it only adds clutter and that information should be available on the changeset. -Alex On Fri, Dec 11, 2020, 5:59 AM Joe <[email protected]> wrote: > My plan for importing: > > I will use Microsoft Building Footprint Data with Bing satellite imagery, > using JOSM > If someone has been drawed manually already, MapWithAI has a feature to > detect if the area > has a building drawed already, and won't download the data repeatly, so > they can keep the work therefore doesn't interfere with these other efforts > > I will check of blatant errors such as the building no longer to exist or > wrongly generated by AI > such as a building outline in farmland that not exist while importing to. > > Area to import: > Any Cities and towns California, United States if the buildings there area > not mapped already is there are to > > data to import: > Buildings shapes that projected to ground, no other information such as > address or height > > There are lucky for mapper who mapping United States is the plenty of > ultra high resolution of satellite imagery, and imagery date checkup in > Bing satellite imagery with JOSM. Most urban area are very new as > within 1~2 years > > Every building object will be tagged with > source=microsoft/BuildingFootprints. Changeset comment will be tagged with > the area such as Town or City, for example : *Imperial County, > California *or *Rural area nearby of Visalia, California * > > Goal: > Goal are following: > Build the map and build community. > To add every building really exist but unmapped in the United States to > OpenStreetMap without creating duplicates. > > https://wiki.openstreetmap.org/wiki/Import/Catalogue/US/BingBuildings#Goals > > Thanks to Microsoft's AI system and satellite imagery provider > > No other datasets will be used with this import. If a building unmapped > and if Microsoft Building Footprint Data has no data, either leave unmapped > and draw manually. Such case, I will add source=manual_draw > > No monetary remuneration will be received of this import > > ------- > > > 11.12.2020, 16:46, "Minh Nguyen" <[email protected]>: > > > > Can you elaborate on your plans? The two pages you linked to are > detailed but intended as only a starting point for various import > efforts. In and of themselves, they don't satisfy the import guideline's > requirement for fully documenting your proposed import on the wiki. For > example, see the full proposals listed in the table at [1]. > > > Data to Import=building outlines > Area=Cities and towns California, United States, if the area not mapped > and existing in reality and able to seen in Bing satellite imagery > source=Microsoft building footprints > > > Can you be more specific? There are a number of completed, ongoing, and > planned building imports in California. [2] It's important that your > import doesn't interfere with these other efforts. In fact, perhaps you > should consider contributing to those efforts, if you haven't already. > (For context, I've helped with the San Jose building import, which is > still ongoing and would welcome help with task completion or validation. > [3]) > > Have you considered any other datasets before deciding to import the > Microsoft building dataset? In California, we're lucky that works by > state and local government agencies are automatically in the public > domain by law. So for many cities and counties, there are likely better > alternatives to the Microsoft building dataset. These alternatives > typically contain attributes that Microsoft's computer vision-based > approach could not obtain, such as addresses and building heights, and > the geometries may be more accurate. If this is the case, I think there > would be a strong argument for preferring the local dataset and only > falling back to the Microsoft dataset in areas that aren't so fortunate. > > Thank you for your consideration and enthusiasm for improving our > coverage of California. > > [1] > > https://wiki.openstreetmap.org/wiki/Microsoft_Building_Footprint_Data#Related_import_proposals > [2] https://wiki.openstreetmap.org/wiki/Category:Import_from_California > [3] > > https://wiki.openstreetmap.org/wiki/Santa_Clara_County,_California/San_Jose_building_import > > -- > [email protected] > > > _______________________________________________ > Imports-us mailing list > [email protected] > https://lists.openstreetmap.org/listinfo/imports-us > > _______________________________________________ > Imports-us mailing list > [email protected] > https://lists.openstreetmap.org/listinfo/imports-us >
_______________________________________________ Imports-us mailing list [email protected] https://lists.openstreetmap.org/listinfo/imports-us
