You are here

Missing photo, visual vs. verbal metadata

Tags: 

A recent comment to a photo in my Flickr gallery: "I've been looking for a photo. I never marked it permanently, and now I fear it's lost forever. It looked a bit like this, and some others you took - it was a horizontal composition, with a lamp hanging from the ceiling of what I assume is a kitchen. There's a wooden table in this early-morning darkness, and to balance the unlit lamp on the left side of the photo, there's a gorgeously lit window looking out into the day. The table is pushed under the window, and on the table are some red flowers that look almost exactly like these, in your photo. The photo was taken in Finland. I really hope I can find it again...

"It was taken in Finland", huh?? Well then! Here, it must be right here somewhere.

But then again - perhaps in a few years we do have such a serious metadata engine that turns visual characteristics into a verbal description like the one above. Not such a bad idea, for someone with a bit of research funding and an inspired team. If they can do that with music -I never thought of this before but why can't we text search inside Pandora with the songs' musical qualities as well as giving the database a referral song?

Thumbsplus - software already has "find similar image" - function. As far as I know it does not add verbal descriptions yet, but who knows what they have on their product development roadmap.

Kommentit

Quite interesting topic to explore. Actually found a few trials on the area:

http://www.tiltomo.com/ -playing on Flickr databases
http://labs.systemone.at/retrievr/ : shape drawing tool based on Flickr
http://grant.robinson.name/projects/montage-a-google/ : Create a montage of similar images.

Fascinating topic to explore...

Thanks for the links, Tommi, I'll have to study those now that I've been lured to take an interest in the issue I didn't know existed.

Kommentoi

Subscribe to Comments for "Missing photo, visual vs. verbal metadata"