It was great to see Masahide Kanzaki again at WWW2005.
He’s been thinking about SPARQL and parts of images, and has put up some cool sample SPARQL queries and XSLT over the results to display parts of images. For example:

Regions with a title containing the word “Dan”:
PREFIX img:
PREFIX dc:
SELECT ?image ?title ?coords ?description WHERE {
?image img:hasPart ?region .
?region dc:title ?title .
FILTER regex (?title, “Dan”) .
?region img:points ?coords .
{?region dc:description ?description .} UNION {?region img:depicts ?x. ?x dc:description ?description .}
}
and even a picture of me as part of a picture from Flickr using his Flickr to RDF converter. Very cool indeed (as ever 🙂
Masahide’s using Andy Seaborne’s SPARQLer demo via a gateway because Firefox doesn’t seem to apply XSLT from different server than the target XML