Image Labeling for Blind Helps Machines 'Think'
From: Washington Post - 11/21/2006 - P. A2
By: Zachary A. Goldfarb

An online game has been designed to make image labeling fun and make surfing
the Internet easier for the blind. The type of Internet program used by the
blind reads Web pages aloud, but since images cannot be identified by the
program and thus have no way of being spoken, many pages are prohibitive. The
solution to this is image labeling, which would give these programs a way to
describe images verbally. Carnegie Mellon University computer science
professor Luis von Ahn developed the ESP game for just this purpose: random
visitors to ESPgame.org are paired up and challenged to provide identical
labels for the image they see; some people have spent as much as 40 hours a
week on the site. Programs such as the ESP Game are known as human
computation, where a computer asks a human a question and the human does the
answering. Teaching a computer using human computation is a lot like the way
children learn to identify things, but as von Ahn says, "Nobody bothers to
teach a computer." CMU's Manuel Blum, who advised von Ahn's dissertation,
explains, "What he's doing is mining the ability of humans." Von Ahn aims to
develop computer intelligence that resembles that of humans and could perform
language translation that accounts for the subtleties of foreign languages,
for example, or make fast illness diagnoses in hospitals. Von Ahn says his
goal is "To be able to use all of this data and to have computers be able to
do pretty much everything we can do."  

Read the entire article at:
http://www.washingtonpost.com/wp-dyn/content/article/2006/11/20/AR2006112001200.html

Links:
Luis von Ahn
http://www.cs.cmu.edu/~biglou/

Luis von Ahn - Research
http://www.cs.cmu.edu/~biglou/research.html

ESPgame
http://www.espgame.org/
