OpenNI is a generic collection of software for interfacing with NUI devices. The project is organized into layers: the hardware drivers, the middleware (gesture recognition, skeleton tracking, etc), and the glue--OpenNI itself. Since components can be dropped in as necessary, the software is extremely flexible but tough to get working.


All Platforms


OpenNI itself consists of three parts:
SensorKinect Driver
OpenNI Library
NITE Middleware (gesture support)

We will also be installing SWIG to allow Python to interface with OpenNI, and OpenCV to allow Python to work with image data. Be sure to get the 32-bit versions of absolutely everything you install for guaranteed compatability.

All Platforms
  1. Download our OpenNI Installer Package, which includes all three components of OpenNI for all platforms. Windows users should run the installers; for other platforms, you'll need to read the install documentation for each component .
  2. Install Python, SWIG, and NumPy, if you don't have them already.
  3. Once installed, run python To track a user, you may need to stand in clear view of the Kinect and make a field goal with your arms for ten seconds in order to calibrate OpenNI.
  1. Get Visual Studio, if you don't have it already.
  2. Download and run the OpenCV superpack installer.
  3. Add the following to your path. (Need help?)
    <Visual Studio Install>\Common7\IDE
    <Python Directory>
    <SWIG Directory>
    <OpenNI Directory>\Bin
    <NITE Directory>\Bin
  4. Download the Windows demo files.
  5. Compile the demo using nmake. Details »

  1. Install OpenCV by following the InstallGuide.
  2. Download the Linux demo files.
  3. Build the demo module using make. Details »


User Guide » | NITE documentation in Installer Package

C++ » | Python »