Thanks to recent developments, actual information technologies offer tools for the access to information about cultural heritage buildings. These tools are aimed to help experts to enrich and share information about historic monuments. This article describes a hybrid process for the semantic annotation of a set of spatially-oriented images (whose positions and orientations in space are automatically estimated). First of all, a large set of images is processed with an automated image based reconstruction method so as to generate 3D information (specifically 3D coordinates). This 3D information is then used in an automated transfer process of annotations through the set of images. At last, new images can be oriented within the set of already-oriented and already-annotated images. Then, annotations can be precisely projected on newly-oriented images. As a consequence, this process provides a simple way to establish a continuity of information inside a set of images.