Paper Reading #20: A Multimodal Labeling Interface...

Comments:

Comment 1

Comment 2

Reference Information:

Title: A Multimodal Labeling Interface for Wearable Computing

Authors: Shanqing Li and Yunde Jia

Presentation: (Conference Paper) IUI '10, February 7-10, 2010, Hong Kong China.

Summary: In this paper, the authors provide a solution for the inconvenience of labeling objects with portable keyboards and mice when using wearable environments. They developed an interface that uses visual and audio modalities which work together in order to achieve the desired results. How it works is that the wearer visually tracks the the object with the integrated camera and the pointing gesture tracking feature, and then using a speech recognition library the user can speak out the label for the object. Besides the gesture tracking system, they also propose a virtual touchpad interface where 

the wearer can identify the object in a more intuitive 

way. 

The evaluation of the system was carried out by encircling several circle regions with different radii and giving it labels. The application given for this interface and discussed in the paper consisted of online learning under wearable computing environments.

Discussion: Even though I think this is a really interesting interface, I think their evaluation methods were really poor. I think this development gives room for some really interesting user evaluations that would of provided researchers with better feedback than what they got. Also, I feel like this system can be implemented for a variety of application and if they would have discussed them in the beginning of the paper it would have made a great difference in, at least my reactions about the paper.