Re: Photo metadata tool

On Thu, 17 May 2001, Yves Lafon wrote:

> On Wed, 16 May 2001, Libby Miller wrote:
>
> Hi,
>
> > I had a bit of a go at something like this, because I was so impressed
> > with the RDFPic tool. Instead of embedding RDF in the images however, I
> > stored all the information, including annotations and more fundamental
> > cataloging information in a database. There's a little demo at
>
> I wanted to have the rdf embedded in the picture to avoid losing comments
> when the picture is transferred, or changed. As the server can, with the
> right Accept: headers, return only the comment (and the same is possible
> for HTTP PUT). Of course it doesn't preclude storing a copy of the
> metadata in a db.
> I like self-describing objects in general :)

Me too :)

	http://www.tasi.ac.uk/2000/09/rdfmeta/
	"RDF for self-describing images"

...something I wrote a couple of years ago, occasionally updated. BTW Dave
Beckett's just been re-packaging his PNG-meta (see
ttp://sourceforge.net/projects/pmt/) that I used in the original demo to
show RDF being stored inside PNGs. Never got around to doing JPEG but then
RDFPic arrived!

My inclination is to manage the RDF alongside the images, then pack it
inside the pics for distribution. I say this in part because often the
metadata you're writing (eg. for travel photos) applies to several files
at once, so it's a tidier management strategy to keep a single .rdf file
for the entire collection, even if it gets copied into each image for
distribution. Either way it would be handy to have a Web interface to
"store" and "extract" services...

My current hobbyhorse is to explore ways of representing (via
annotations or in the original metadata) that a picture
*depicts* some individual or individuals, so we can build fun tools that
connect people visually by finding paths through the graph.

There are some examples of this in RDFWeb:

http://rdfweb.org/ ->
http://rdfweb.org/~pldab/rdfweb/danbri-foaf.rdf

excerpt: [[  ...
<knowsWell>
        <web:Description about="genid:emiller">
        <mbox web:resource="mailto:emiller@oclc.org"/>
        <img web:resource="http://www.oclc.org/~emiller/capture.jpg"/>
        <depiction>
                <web:Description   about="http://www.ilrt.bristol.ac.uk/people/cmdjb/events/dc7/orig/eric.png">
                <dc:title>Eric Miller, relaxing.</dc:title>
                <dc:format>image/png</dc:format>
                </web:Description>
        </depiction>
        <homepage web:resource="http://purl.org/net/eric/"/>
        <rdfs:seeAlso  web:resource="http://purl.org/net/eric/webwho.xrdf"/>
        <name>Eric Miller</name>
        </web:Description>
</knowsWell>

]]

In this example, I say in RDF that some person is depicted by some image;
for image annotations that's kind of back-to-front, but the principle is
the same.

Dan

Received on Thursday, 17 May 2001 10:30:16 UTC