A caption annotation scheme for georeferencing images

Abstract

We propose a caption annotation scheme to address the task of obtaining a geo-footprint for an image from caption text, introducing tags to identify image subject and using SpatialML to annotate and ground references to image subject location.