WikiProject Belgium/open streetlevel imagery

From OpenStreetMap Wiki
Jump to navigation Jump to search

Open StreetLevel Imagery project

The main description of this project is on the website. This page is for getting into more detail.

We are always looking for volunteers to use one of our cameras. Send us a message at on this topic!

You can find a starter's manual here.

Our choice of set-up

After some research, we decided to go for the GoPro Max. Most of our images are taken on a bicycle, some with a car.
Bicycle shopping list:
- GoPro Vented Helmet Strap Mount
- Hama Helmet mount front
- RAM suction cups (RAM-B-189B-PIV1U) with a variety of extensions depending on the car.
- charger and two extra batteries (for a full day of autonomy)
- SD cards. Preferably with high writing speed and high capacity (we also use the SD cards to send backups by post). Also an adaptor for regular SD card format (as many laptops have an SD card reader)

We use the TimeLapse mode, which takes an image every 2 seconds. This has the advantage of having zero post-processing. Disadvantage is that there in no horizon leveling and a relatively low frame rate.

We have a very short manual to get you started (NL). TODO: translate and put on the wiki.

Using the images for mapping

The resulting 360° coverage is visible here (blue what we collected, green by other people).


Osmose downloads traffic signs that were detected with the imagery and turns them into microtask. If it sees a sign "max allowed speed is 30", and it finds no nearby road with maxspeed=30 in OSM, it will create a microtask.

You can see these issues with this direct link. Known bug: depending on the day, only issues in Wallonia and Brussels are shown, or only in Flanders.

To fix an issue, simply click on "id-zone" to open it in the web editor, or "josm-zone" to open in JOSM. In iD, use the "Map data" button on the right to activate the Mapillary traffic signs. The detected signs will show up on the map. Click on a detection to see the image it was seen on. If you want to see more images of the area, you can open the Mapillary webapp in the area by clicking on the logo in the bottom right of the image.

There's a lot of false positives. Simply mark them with the red X in Osmose, and they will disappear from the map. Common errors in the detections:
- wrongly interpreted sign (e.g. maxweight read as speed)
- temporary signs during works
- sign could have been removed (check the date of the detection shown in Osmose)

Tasking Manager

To work through the images systematically, the tasking manager is useful. For example, once coverage of Pepingen municipality was complete, was used to systematically check all the streets. For speed limits, having full coverage is really useful, because missing a single zonal sign can lead to huge differences. This kind of work is easiest when combined with a nice overpass rendering or a MapComplete theme of the topic. Even with full coverage, there can still be errors in the traffic signs - which you can map in the VerkeerdeBordenDatabank.

If mapping in Flanders, consider adding the official database of traffic signs to your editor:{proj}&WIDTH={width}&HEIGHT={height}&BBOX={bbox}. You can also view them on the Geopunt website.

This kind of approach might also be useful for maxweight/maxlength/maxheight mapping.

Download the data

It is possible to download detections through the Mapillary API. More straightforward is going to the Mapillary webapp, activating the signs or objects you want and downloading them. This can only be done for relatively small areas though. For example, this link should show most traffic signs related to maximum speed. Click on the "map data" button on the top left of the screen and click download to get a GeoJSON. This can then be opened in JOSM or a standard GIS software.