![]() |
New Google Image Labeler project is idiot-prone.
|
that was kinda lame and what are these points for? i quit after the first photo
|
They really know their weakest link in search. Their current image search is useless almost. To find out ppl's image labeling pattern, they can improve upon.
|
Quote:
So given this picture, http://www.diseno-art.com/images/ferrari_250_GTO.jpg Most people will just go. Red, car, wheels, old, holes, black. As opposed to "red car", "black background", "old car", or "wire wheels" despite the prompt asking you to be specific for more points. (The correct answer is Ferrari 250 GTO.) I think this can make an image search bloated and too vague. Of course, Google's not going to trash its current system (which I think is fine) for this but use this to supplement the search engine. In theory it works. Make people label individual picture so that it's easier for people to find the right pictures when they need to. But Google took this approach in a completely wrong way. (like an open wikipedia page) And they weren't specific enough on what they want people to do with the labeling. They'd great if you're looking for a stock photo, but useless when trying to find specific photos. It's still a beta so. We'll see. |
Good point, Heil.
I discovered exactly what you were saying in the first picture it brought up. My partner described a picture of "clouds" as "Blue, white, sky, fluffy". ..Fluffy? Come on. I can't see how the majority of these users are going to help narrow done the relation of words to pictures with answers as ambiguous this. |
I will name the picture red sex in the dark.
|
| All times are GMT -4. The time now is 05:15 PM. |
Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2026, vBulletin Solutions, Inc.