Map features represent objects that Mapillary has detected in images and placed in a 3D space. Using computer vision, we recognize objects captured in images (available as object detections). Next, by reconstructing the 3D scene structure, we can position these objects on the map as map features.
Mapillary extracts 42 types of objects as point features (i.e. map features that are represented by points on the map, like fire hydrants or street lights), in addition to 1,500 different traffic sign classes (also extracted as point features). 19 object classes are extracted as line features (map features represented by lines, such as guardrails or lanes).
To retrieve and use automatically extracted map features, you need a subscription through Mapillary for Organizations. Create a Mapillary account and set up an organization to get started. Note that line features are currently not available under a commercial license.
Characteristics of map features
When retrieving map features as a dataset, you will get the type of the map feature and its GPS position (latitude and longitude) in a standard geospatial format (shapefile or GeoJSON).
There are two important notes on map features.
- To estimate the location of the object on the map, the object needs to be detected in at least two (for traffic signs) or three (for other objects) images. The positions of the images are then used to calculate the position of the map feature. Keep this in mind when choosing the interval of the images you capture—more images means more data points to triangulate the position of the object more closely.
- The accuracy of the location of the map feature will be influenced by the accuracy of the location of the images. This is contingent on the GPS accuracy of your phone, camera, or GPS device. If high-accuracy positioning is a requirement, you might consider professional devices with enterprise-grade GPS capabilities.
Point features are map features that can be marked on the map as points, such as trash cans, traffic lights, and manholes.
Mapillary extracts 42 types of objects as points. That means, we provide the estimated coordinates of the objects, based on the positions of the images that we've detected the objects in.
The following objects can be extracted from Mapillary imagery as point features:
|- Temporary Barrier
- Crosswalk - Plain
- Lane Marking - Arrow (Left)
- Lane Marking - Arrow (Right)
- Lane Marking - Arrow (Split Left or Straight)
- Lane Marking - Arrow (Split Right or Straight)
- Lane Marking - Arrow (Straight)
- Lane Marking - Crosswalk
- Lane Marking - Give Way (Row)
- Lane Marking - Give Way (Single)
- Lane Marking - Other
- Lane Marking - Stop Line
- Lane Marking - Symbol (Bicycle)
- Lane Marking - Text
- Bike Rack
- Catch Basin
- CCTV Camera
- Fire Hydrant
|- Junction Box
- Parking Meter
- Phone Booth
- Signage - Advertisement
- Signage - Information
- Signage - Store
- Street Light
- Traffic Sign Frame
- Utility Pole
- Traffic Cone
- Traffic Light - Cyclists
- Traffic Light - General (Horizontal)
- Traffic Light - General (Single)
- Traffic Light - General (Upright)
- Traffic Light - Other
- Traffic Light - Pedestrians
- Trash Can
- Water Valve
Take a look at the articles about exploring point features on the Mapillary web app.
Traffic signs are also point features but presented separately from other objects on Mapillary because they are further categorized into more than 1,500 traffic sign classes from 100 countries. That means, we provide the estimated coordinates of the different traffic signs (such as stop signs or speed limits) based on the positions of the images that we've detected the signs in.
Line features are a map representation of line-shaped real-life objects, such as lane markings, rail tracks, and curb cuts.
Line features are matched against existing map data. The features currently available on Mapillary have been aligned with the OpenStreetMap road network but our processing system has been built so that it can utilize any map data to match against. This means we will be able to extract line features aligned to any road network data in the future.
Take a look at the article about exploring line features on the Mapillary web app.
The following objects can be extracted from Mapillary imagery as line features:
- service lane
- traffic island
- dashed lane marking
- solid lane marking
- bike lane
|- curb cut
- rail track
- road shoulder
- other barrier