Wednesday, February 13, 2008

Google Image Labeler OR It's Like Crack

I watched UFC 81 tonight, which was waiting on Rich & Lisa's DVR since Saturday night.  I was out celebrating Tricia's birthday with a group of her friends, and her man Garrett.  I dragged them to the jazz club, where we stayed until 5 AM or so.  Will I never learn?  I was dragging the next day.  We lost our hockey game Sunday, and I was victim to some cheap shots by their defense men in front of their net.  When the ref did call something, they would try to instigate something to get a retaliation to try and get a matching penalty.  Very cheap, especially when they were winning.

My latest addiction is the Google Image Labeler.  It's so random and engaging, it's usually more exciting than actually doing my work.  Try it.  It's like a combination of Taboo, $10,000 Pyramid, and a Rorschach inkblot test, but with random web images.  It pairs you up with a partner for two minutes and you start entering keywords to describe the image which appears.  Previously chosen keywords are off limits. When you and your partner have both entered a matching keyword, it goes to the next image and you are awarded points based on the relevance and quality of the keyword.  If you run out of ideas, you can click "pass" and go to the next image.  If you sign into Google, it will keep a running tally of your points.  I'm embarrassed to say how many points I have.  Think 7 digits.  You start with the obvious.  If it's a picture of people, you just type people.  If there are words in the picture, you can type the words.  If the image is from a movie, the word movie or show generally matches up with your partner.  My problem is recognizing celebrities and older models of cars.  I'm sure someone on the other end who has entered 11 keywords is screaming at me when I'm clicking the "pass" button and waiting for them to do the same when it's somebody obvious.  I just don't read People or watch E!  Sorry.  And I don't read Car and Driver, either.

No comments:

Post a Comment