People and organizations across the globe share their imagery in the world’s largest publicly available database. Powerful search tools help you find the time, place, and objects of interest to get a street-level view and do virtual inspections.
Scale and speed up map updates and asset inventories. Using computer vision, our technology recognizes objects in street-level imagery and places them on the map. Support for over 40 object classes and more than 1,500 different traffic signs.
Objects automatically detected in images
Objects located on the map as point features
Global traffic sign recognition
Automatic object detection
More than 60 object types automatically detected in imagery by award-winning machine learning algorithms for street scenes.
See which objects we detect
Point features on the map
Estimated coordinates of objects (triangulated from the images where the object has been detected). Currently supported for more than 40 object classes, including a further classification of traffic signs.
See which map features are available
Pixel-wise manual annotations in the Mapillary Vistas Dataset
Bounding box annotations in the Mapillary Traffic Sign Dataset
Machine-labeled images from the Mapillary database
Training data for machine learning
Train models that understand street scenes around the world to boost autonomous mobility and transport. Develop semantic segmentation models with the Mapillary Vistas Dataset and traffic sign recognition models with the Mapillary Traffic Sign Dataset, or pick a custom imagery set from our database.
Mapillary Vistas Dataset
A benchmark dataset of manually annotated training data for semantic segmentation of street scenes. 25,000 images pixel-accurately labeled into 152 object categories, 100 of those instance-specific.
Learn more about the Mapillary Vistas Dataset