In a method for increasing accuracy of augmented reality information presented to a user navigating a real-world mapped environment, a specific object within a field of view of a camera of the user's mobile device is identified, at least by predicting an object type by a machine learning model processing images captured by the camera, and identifying a node that corresponds to the object type within a map database representing the mapped environment. A location associated with the node is retrieved from the map database, and is used to update or confirm one or more estimated locations of the mobile device. Digital information is caused to be superimposed on a real-world view presented to the user via a display of the mobile device, the digital information being selected based at least in part on the updated or confirmed one or more estimated locations of the mobile device.