Notebook
April 11th, 2005
Categories: Amazon, Google, Interaction

Combining the Amazon Yellow Pages‘ street-level images of businesses with Google Maps:

  1. The info balloon for a “Tula’s, Seattle WA” search result.
  2. Clicking the left arrow several times moves up the block to Mama’s Mexican Kitchen.

If you haven’t seen it already, Robin Sloan’s future history of mass media, “EPIC 2014” is somewhat entertaining.

May 16th, 2004
Categories: EXIF, Google, Meta-data, RDF

Continuing my EXIF jag…

One way Google could present EXIF meta-data for searching.

May 13th, 2004
Categories: Browsers, Google, Interface

For image contextual menus, browsers could present a “Search Google for Similar Images” command, with the query constructed from the image’s filename, alt and title text (conjunctions would be removed), and perhaps the page’s meta keywords.

EXIF meta-data would be an even better source of information upon which to base such a query and would probably deliver more accurate results.

May 8th, 2004
Categories: EXIF, Google, Meta-data

The ever-useful Google Image Search could become quite a bit more powerful if EXIF meta-data were available and cataloged for use as query factors.

The current advanced search features are good, but fairly general.

With good EXIF support, you could construct an image search with the following parameters:

  1. Taken in the last ten days
  2. Portrait oriented
  3. At least 500 pixels wide
  4. Photographer is Annie Leibovitz

Additional flexibility would be available once all digital cameras have GPS built-in (call me optimistic). Using location-coordinate mappings, you could specify all of the above possibilities along with where the photographs were taken. This wouldn’t be really useful for Leibovitz because she is primarily a portrait photographer, but it would be great for locating images in which the geographic location is important, such as nature scenes and current events.

Verifying that an image was taken by a particular photographer presents a problem. I suppose the source image server could be factored in as one way of measuring accuracy - if a photo came from annieleibovitz.com (just an example - there is currently nothing there), there is a good chance that the EXIF data is correct. If that site is frequently linked to by other sites, that would also lend some credence to the claim.

Currently, EXIF data is usually removed to decrease image file sizes or simply nonexistent, but that probably won’t be the case in the relatively near future as digital cameras entirely replace film for general use (around the time the term “digital” is no longer used when referring to them) and high speed connections become ubiquitous. Building new Image Search features for the future seems like a sound idea.