Improvements on automatic metadata generation and access systems for digital photographers
I have a lot of images online (roughly 16,136 as of April 2, 2003 and now
23,321 as of February 1, 2004).
It's not actually that many images compared to people who might do
this for a living or people who choose to use automatic capture
systems, but my collection has become unmanageable using standard
techniques. I've used various database-like systems in the past to
catalog things, but I'm getting to the point where I'd like to explore
some novel interfaces for using and automatically generating metadata
for images. Right now my personal efforts are on hold as I work with the
Family Video Archive in Gregory Abowd's Ubiquitous Computing research group.
This page is incomplete, but should be updated soon.
Scenarios which I feel are important There are lots of good
papers/descriptions of how image metadata should work, but many of
these come from the library sciences or GIS fields. These are a few
scenarios which are important to people like me who want to maintain
personal media collections and are picked because they somehow
elucidate a difference between the goals of a large-scale digital
librarian and a personal image collector.
- automatic gathering and addition of location data for images
- inference of semantic location data from coordinates
- accelerated tagging through pattern recognition
- and many more...