Jimmy Ho
Evin Schuchardt
Reference Information
Jeffrey P. Bigham, Chandrika Jayant, Hanjie Ji, Greg Little, Andrew MillerĪ³, Robert C. Miller, Robin Miller, Aubrey Tatarowicz, Brandyn White, Samuel White, and Tom Yeh, UIST’10, October 3-6, 2010, New York City, USA
Summary
The main point to this article is on a program that is being developed called VizWiz. The purpose of which is to help specifically blind people answer simple visual based questions for them. The blind user will take a picture of the environment that they have a question about and then speak a question about it. This is then sent off and answered by an actual human and then sent back. The article is generally about how the system is used to get the feedback to the person. Also, the authors deal with the fact that the images are not going to be very good because they are coming off of a cell phone camera. This also is compounded by the fact that the users are blind and therefore can not necessarily frame up a good shot of what they are asking about. This is where the human element helps because humans can see patterns and identify objects much better than a computer could.
Discussion
I like the idea of this program but only strictly in the since of a temporary solution. The whole argument for using human workers was that the computer could not do as well as a human could. However, this does not mean that computer programming will not improve to the point of being able to analyze images just as well as human do. This is the goal that needs to be achieved rather than just bad mouthing how the computer currently cannot do this. In fact, Google has their "Google Goggles" that is very similar to this article. However, I find their research in this area and using humans interesting especially when they say they can actually keep it cheaply enough to be profitable.
No comments:
Post a Comment