The Navigation Data team at Mapbox is releasing 184k turn restriction detections located across 35.2k intersections and 23 cities for the OpenStreetMap community.

Interactive Map

GeoJSON File

We generated these turn restriction detections by applying our machine learning computer vision models to Microsoft Streetside Imagery, from which Mapbox has acquired the right to contribute these detections to OpenStreetMap.

One thing to note is that these “detections” are merely proposed turn restrictions and are not equivalent to substantiated turn restrictions. The mapping team at Mapbox has been manually verifying these detections and subsequently mapping verified turn restrictions onto OpenStreetMap. We are releasing these detections to the OpenStreetMap community, so that those who wish to move faster on improving OpenStreetMap based on these detections can do so.

Microsoft has just integrated its Streetside imagery for the United States into iD, a popular web-based editor for contributing to OpenStreetMap. This is the same imagery currently visible on Bing Maps now embedded into a popular editing application initially developed and now maintained by Mapbox. For each of our turn restriction detection (download the GeoJSON file by clicking on “GeoJSON File” link above), we’ve included:

  • “label” : type of turn restriction detected
  • “prediction_date” : date on which we ran our detection model on Streetside image
  • “bubble_id” : id associated to a streetside “bubble”, or the discrete 360 degree view of a location.
  • “streetside_url” : URL to the Streetside image associated with each detection.

Below are a few examples from the Streetside Imagery API:


Comment from urbigenous on 15 June 2018 at 12:54

This is amazing work and a great application of machine learning. 👏👏👏

Login to leave a comment