Image annotation is a very interesting Semantic Web topic, because it suggests a way in which Semantic Web-based searching could be much more effective than text. There are many different ways in which images can be described - from geographical location to colour to the people depicted to a story told by the image. RDF offers the possibility of describing images in some or all of these dimensions and then enabling searching of image collections by these different facets and others.
Work within SWAD-Europe on image description has had several strands. Two developer workshops have been held on aspects of image description. The first focussed on surveying and demonstrating tools. The second was mostly about the vocabularies used and the some of issues with combining them together.
Also as part of the dissemination strand of the project, SWAD-Europe has participated in community efforts to create vocabularies for image description. The first was W3Photo, a project to annotate photos from the WWW series of conferences. Also involved were Greg Elin, the Mindswap group from Maryland, the IAM group from Southampton, and others including Jim Ley, Morten Frederiksen, Masahide Kanzaki and Benjamin Nowack.
From the point of view of the SWAD-Europe project, our interests were:
Information about the meetings held is on the ESW wiki WWW2004; the specification is at W3PhotoSpec, the vocabularies used are documented at: W3PhotoVocabs, including the vocabulary for describing parts of images, which is hosted at Benjamin Nowack's site.
There are several tools available for annotating parts of an image, for example Jim Ley's SVG tool and RDFPic as demonstrated at the SWAD-E image annotation workshop; also Greg Elin's fotonotes, Masahide Kanzaki's javascript image annotator, and a java tool called Photostuff from the Mindswap group.
Discussions in W3Photo centred around the requirements for a vocabulary for describing parts of images, which would include
Discussions about mixing vocabularies for image annotation have sometimes occured on the FOAF mailing list as part of the codepiction project and similar efforts by independent programmers. They have often centred around adding geographical information to photos, but also mixing in event information and Dublin Core data, as well as the core 'depicts/depiction' information about who or what is depicted in the photo, using FOAF and Wordnet classes and properties. The w3photo project is also concerned with rights and licensing of the metadata and images for use as sample data for different systems.
Further discussions were held at the Foaf Galway SWAD-Europe final workshop about a basic set of properties and classes specifically for photo annotation. This time we steered clear of the complexities of describing parts of images and focused just on some very basic useful properties: date, place, event, depiction. Notes from the discussions are available on the ESW wiki. The requirements in this case were that
Work continues in this area with discussion on the rdfweb-dev list.
As part of SWAD-Europe, a simple tool for photo annotation was created which enabled this sort of simple photo annotation of online photos. It uses existing online RDF data sources so that the annotator can search for people, locations and events and reuse information where available.