SQLServerCentral is supported by Red Gate Software Ltd.
 
Log in  ::  Register  ::  Not logged in
 
 
 

SQL Musings

Add to Technorati Favorites Add to Google
Author Bio
Steve Jones Editor at SQLServerCentral.com You can follow Steve on Twitter as way0utwest (www.twitter.com/way0utwest)
Browse by Tag : Bing (RSS)

Searching for Data in Images

By Steve Jones in SQL Musings | 09-22-2009 6:39 AM | Categories: Filed under: ,
Rating: (not yet rated) Rate this |  Discuss | 732 Reads | 96 Reads in Last 30 Days |4 comment(s)

I tried out Bing’s new visual search after seeing a few people note it on Twitter. I was a little excited to try it, thinking that it might help me to find images for the editorials easier.

It didn’t, and I was disappointed. I thought that this short review in PC Week summed it up. It just doesn’t work well, and the best quote:

Yet after every search trial I always come back to Google, because it works.

I tried typing something in, but that brought me back to the regular search. I clicked “TV Shows”, looking for images from Leverage, but I couldn’t make out detail in the images. Too many to look through, and too slow. I found myself looking up at the search box for the words as I moused over

I’ve given Yahoo and Live chances, and each time they just don’t do a better job. So why should I switch? I haven’t used Bing a lot, but unless it improves on Google significantly, why should I switch?

What I want

I’d use a visual search if it made things easier. What I’d like in visual search is a combination of meta data and breadcrumbs to narrow things down. If I search for “flower”, then give me some options on the side like Amazon does. Let me pick from

  • Media (books, movies, pictures)
  • Red, pink, white, red, or even “more red” or “less red” (or other colors)
  • natural, in stores
  • people

Then let me click those to narrow down the images I want. I can scan images quickly if I’m trying to narrow things down or find some type of image.

If someone could make search for better, I’d try it, but so far Google does the job.

(cross posted to my dkranch blog)