What You Need to ProceedĪll you need to try out this basic HMM-based gesture recognition technique for yourself is a working install of GNU Octave (which means Matlab should more or less work, as well). But in this article, we’ll simplify the task, use some pre-existing machinery as a subsystem for our code, and attempt to perform–or at least understand–simple gesture recognition on data from the Kinect. The answer can quickly lead us into deep computer vision and machine learning territory that will send noobs packing. Now, suppose we want to do this kind of analysis ourselves. Though depth cameras are not exclusive to Microsoft (and, in fact, the Kinect’s hardware design is licensed from the Israeli company, PrimeSense), Microsoft’s novel use of the device as a game controller is driven by the company’s own software for analyzing its 3D data, including proprietary algorithms for feature selection, scene analysis, motion tracking, skeletal tracking and gesture recognition. Microsoft’s Kinect sensor is the first large-scale, commercial release of a depth camera device-a camera that can see in 3D. I also worked with my friend Frederik Lang on the material presented here, but the shortcomings remain all my own. Note: The precious little I know about machine learning I learned from the brilliant Frank Wood, whose talk at the Machine Learning meetup inspired me to finally put together this post.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. Archives
January 2023
Categories |