Comment 1
Comment 2
Reference:
A $3 Gesture Recognizer - Simple Gesture Recognition for Devices Equipped with 3D Acceleration Sensors
Sven Kratz, Michael Rohs
IUI '10
Summary:
This paper examines a simple, easy to implement 3D gesture recognition algorithm that can be incorporated into any device with 3D acceleration sensors. It requires no library support or extra hardware, and it is intended to be a tool for user interface prototyping. The algorithm continuously samples changes in acceleration and performs a summation of the data. This data is then compared to a table of training gestures previously recorded by the user. A heuristic is applied to find the highest scoring candidate in the table, and the closest match is selected. Based on initial testing, the $3 Gesture Recognizer has an average success rate of around 80%, while individual users scored anywhere between 58% and 98% correct gestures recognized.Discussion:
While this algorithm would perform poorly in a commercial product, for prototyping and testing uses it seems ideal. It can be used to examine the limitations of any given 3D motion enabled hardware, test a proof of concept, or simply explore what gestures work better than others before committing to an elaborate, expensive control scheme. It may be quick, dirty, and not 100% accurate, but it gets the job done for the task it was created for.
As someone who's tinkered around with random electronics prototyping, being able to do these sort of motion gestures would be really interesting to experiment with.
ReplyDeleteThis could be a really cool tool to mess with and include in some simple devices to see how it worked. The simplistic nature of it only uses mathematical functions was pretty impressive.
ReplyDelete