Program your computer to detect custom gestures for your specific needs! The Kinect Custom Gesture Recognition Program allows users to input in detail and label personalized gestures. This video by Michael displays the program at work and how the program can be beneficial to other utility programs. In the video, the user displays how a gesture can be registered into the program. Specific gestures of the user is tracked by the Kinect on key joints. The program remembers the joints and the flow that led to it. The user then can label these gestures, allowing the program to display the label later on. Even if the user forgot a gesture, he can check the his joint reply. This is indeed a great foundation program for devices that will be needing gesture recognition for either entertainment, use or security.

Below are the list of functions that the program has:

“Here are some functions:
-Search for multiple gestures.
-Disable or priotize gestures.
-5 gesture types to support almost all gestures: Stateless, Statefull, Directional, Bidirectional and PositionsOnly
-Customizable tolerance value and return range of the values.
-Define which axis are important for the gesture and which can be ignored.
-Define which axis are relevant for the return value.
-The Orientation of the user can be set, to be ignored, so he can rotate around.
-the joints which the gesture should include can be selected.
-Gestures can be found on every user which is in the scene, but also users can be ignored.
-Automatic calibration of the users.
-Create a gesture automatically by moving the joints, or by setting poses (like showed in this video).
-All gesture settings can be changed, after a gesture has been saved.
-Also the joints of the gesture can be changed, after is has been recorded.
-The joints of a gesture can be shown in the 3D engine, while the user walks around.”

For more information about the Kinect Custom Gesture Recognition Program, visit the project’s Youtube Page.

Visit Website


  1. Just an information, if possible. For the gesture recognition part, did you use some feature of NITE, or only OpenNI (as suggested from the title of the video 😀 )?
    However, congratulation for this great work!


Please enter your comment!
Please enter your name here