Thursday, April 14, 2011

Paper Reading #22

Comments
Patrick Frith
Michael Atkinson

Reference Information
Title: A $3 gesture recognizer: simple gesture recognition for devices equipped with 3D acceleration senors
Authors: Sven Kratz, Michael Rohs
IUI 2010




Summary
   The concept of this paper was a system with a simple gesture recognizer. This recognizer uses a 3D accelerometer as an input from a Wii controller or some other form of mobile device. The system has an algorithm which is made to be independent of devices. It uses geometric calculations as well as trigonometric ones. This system was tested using a Wii controller. Everybody in the study tried particular gestures. The study showed that the system was not as well off as some of the more complex ones.



Discussion
    I enjoyed reading this paper. The concept seems really cool. I do not see how applicable this particular system could be in today's world. The fact that it is limited to only about 15 gestures just seems like it would limit the system. I think the first step in improving this system is to increase it's ability to recognize more gestures.

4 comments:

  1. I agree that the system seems very limited, but since they were just targeting it at prototype systems, it seems like it wouldn't be a problem for that purpose.

    ReplyDelete
  2. This is kind of cool, but it seems like the Microsoft Kinect is much more natural to use.

    ReplyDelete
  3. I agree that the small amount of gestures is a problem. It seems like most programs that use gesture recognition have some issues with precision.

    ReplyDelete
  4. I suppose that because they were intending their development for testing prototypes, their precision and recognition issues are not that relevant. However, it was nice to have them explain that in the paper, since when you first read their limitation, you may think that there is nothing good about the system.

    ReplyDelete