Types of map data

Mapillary’s computer vision technology automatically detects objects depicted in images, such as buildings, cars, pedestrians, traffic signs, bicycle racks, and much more. This is based on a method called semantic segmentation—an algorithm is trained to detect and assign a category label to every pixel in the image.

Another part of our computer vision technology deals with 3D reconstruction of places from the images. By connecting object detection and 3D reconstruction, we are able to triangulate the location of an object that has been detected in several images. That way, we can determine the location of that object in the real world and place it on the map.

Currently, some of our detections are available as map features, while others are not yet. But it's still possible to extract them object labels together with the coordinates of the images that they were detected in. This means that altogether, you can get two kinds of map data from Mapillary: map features and object labels. Our map features are one of two types: traffic signs (as point features) and a number of line features (currently in beta).

To recap, this is what we offer:

Object labels—detected objects in imagery provisioned with the image locations (not triangulated as map features). Read more.

Traffic signs—detected traffic signs in imagery, triangulated between multiple images and extracted as point features on the map. Read more.

Line features (beta)—detected line-shaped objects in imagery, aligned with the road network and extracted as line features on the map. Read more.