Dniepropetrovsk Buildings Import

In October 2015, M@ster_dp approached Ukrainian OSM community and offered to import building information of city Dniepropetrovsk which was produced by his company. He is the legal owner of the data and gave explicit permission to publish it under ODbL/CT

Signup to be a part of the import https://osm.splashthat.com

The Data

The data contains building outlines, addresses and number of levels for vast majority of buildings in the city.

The original data dump in SHP format, as well as the OSM conversion, could be found here

Number of buildings: 139122

Number of addresses: 87427

Github: https://github.com/osm-ua/dnipropetrovsk-addresses-import

The Plan

Since the city is relatively well mapped, and that building outlines for private houses seem to be better in OSM, there is a need to merge the data.

The amount of work is pretty tremendous, since we have to perform manual checks, and the data has some drawbacks, such as the alignment is sometimes up to 10 meters off. Several persons are going to be involved and we plan to coordinate the effort with the use of OSM Task Manager.

Also, for the copying the tags and shapes over we're going to use a custom JOSM plugin written by Sev.

For easily seeing the buildings without addresses, a JOSM stylesheet needs to be added:

way[eval(JOSM_search("building -\"street:id\" -(child \"street:id\")"))] {
    width: 4;
    color: red;
}

The step by step instructions:

1. Pick up a pie slice and load it in JOSM
2. Load the data file as a separate layer (import layer)
3. Open Bing imagery and adjust it according to the GPS tracks
4. Adjust the selected slice according to the imagery (there are no roads in the dataset, thus we need this 2 steps procedure)
5. Switch to the special mode ('t' or 'Tools' → 'Copier')
6. If the building is present in OSM and shape is OK, but is highlighted in RED (no address), we need to copy over the tags:
   • If the shape on the import layer has common points with the building, e.g. there is an intersection. Click on the common point
   • If the shape on the import layer is nearby. Ctrl+Click on the closest point
7. If building in the underlying layer is better. Shift+Click to copy both tags and shape

This will let to quickly copy the data from the import set.

Progress

2015.10.21 M@ster_dp opens the topic on the forums with the proposal to import

2015.10.22 M@ster_dp clarifies the data source

2015.10.27 First data export is published and tested

2015.11.02 Full dataset is published

2015.11.03 M@ster_dp clarifies that it is ODbL license

2015.11.04 andygol creates a pull reques on OpenAddresses

2015.11.15 Created the first draft of this page

2015.11.21 andygol created the map to compare existing data and data available for the import

created layer for adding it into iD, JOSM - tms:https://api.mapbox.com/v4/andygol.8218824a/{z}/{x}/{y}.png?access_token=pk.eyJ1IjoiYW5keWdvbCIsImEiOiJjamxxbzdzejMyYjNiM3BwYjd6NDh6dmd6In0.ZQtHwTgZoKCHxP1k5ECjXA

2016.01.14 andygol created video describing steps for import data - https://youtu.be/8QJNjmn3o7I

2016.02.10 Mail to imports@ was sent by _sev