What’s the accuracy of the data I get out of Mapillary?
For labels: Mapillary’s traffic sign recognition algorithm has been benchmarked at over 90% for precision and recall. Our semantic segmentation models are ranked #1 for road scenes. For more specific information about accuracy by label type, please refer to our latest research.
For map features: the accuracy of the positions of map features is based on the quality of the 3D reconstruction of the location. This largely depends on the GPS accuracy of the device used to capture the imagery and the density of images in the area. Since Mapillary’s database contains imagery from diverse sources and devices, the accuracy of the position of map features can vary across locations.
If you are capturing your own imagery, use high-resolution cameras with the highest-accuracy GPS receivers within your budget and capture images as closely as possible to maximize accuracy.
In which format can I get the data?
You can get data extracted from Mapillary in geojson or shapefile format. The data file can be pulled directly from your organization dashboard in the Mapillary web app or by using the Mapilary API.
How do you calculate the km in my shape of interest?
We calculate coverage in your area of interest (in kilometers or miles) based on OpenStreetMap road network data. You can review Mapillary coverage in your area of interest via your organization dashboard in the Mapillary web app.
How much coverage is there in my area of interest?
You can review Mapillary coverage in your area of interest via your organization dashboard in the Mapillary web app.
My area of interest has too few images, how can coverage be increased?
If Mapillary coverage does not meet your needs, you can easily capture your own imagery to increase coverage.
Can you capture imagery for me?
Yes, we have a contributor network that captures imagery around the globe. Capture projects are set up on a case by case basis. Please contact firstname.lastname@example.org with your requirements.
Can I upload as many images as I want in the same place?
Yes, additional imagery in the same location does not add to the kilometer or mile count. More data in your area of interest improves the quality of data extraction and unlocks the ability to review changes over time.
When do I need to pay for Mapillary?
The Mapillary platform is free for anyone to use, and all public imagery is available as open data under the CC BY-SA licence. That means you need to share and attribute any derived work under that licence too—e.g. maps or geospatial datasets that have been created or updated using Mapillary imagery as a reference.
If CC BY-SA doesn’t suit your use case, you need a subscription to get a commercial licence for the imagery and derived metadata. You also need a commercial licence for the data that we extract automatically, i.e. detected object labels and triangulated point features.
You don’t need to pay for imagery that you’ve contributed yourself. You (or your organization) are the owner of this imagery and can use it any way you want also on the Mapillary platform.
If you want to use Mapillary imagery (whether it’s all public imagery or just your own contributions) in ArcGIS integrations, you need to purchase a license for Mapillary for ArcGIS.
I want a one-time data pull, do I have to get a subscription?
Our product is offered as subscriptions to reflect the changing nature of the images and data in each area of interest. The prices include updates in the imagery and data during your subscription period, as more images are added and more data is extracted (or as Mapillary’s computer vision algorithms improve). If you don’t need the updates, you can opt to pull the data just once during your subscription period.
For processing services, do I pay for all of the images or only the ones you blur/detect features in?
We charge for processing services based on the total number of images. Our algorithms are run across your entire dataset.
We believe that there is a need for an independent and neutral provider of street-level imagery and map data, untied to any particular mapping platform. We are committed to building a global service for everyone.