DepthJS allows any web page to interact with the Microsoft Kinect using Javascript.
Pull request Compare This branch is 33 commits behind doug:master.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


DepthJS is a browser extension (currently Chrome & Safari) that allows the Microsoft Kinect to talk to any web page. It provides the low-level raw access to the Kinect as well as high-level hand gesture events to simplify development.

Current Status

Moving to OpenNI/NITE based backend, forsaking OpenCV for now. Gesture recognition is thus far better than what we had before. Finger-based gestures will soon follow as a few projects parallel to DepthJS will merge in coming months.

Current gesture language:

  • Wave to start hand tracking and get "blue pointer".
  • Push to click.
  • Circle to end tracking and remove "blue pointer".

Note: OpenNI & NITE should be downloaded and linked appropriately to your build.


DepthJS is very modular. The Kinect driver and computer vision are written on top of libfreenect & OpenCV in C++. This component can output the raw RGB image, the raw depth map (filtered for the hand), as well as the high-level events that the computer vision recognizes. A native browser plugin (think Flash) wraps this Kinect code, which directly interacts with a Javascript browser plugin. Fortunately in Chrome extensions can contain native code, so it will be easy for anyone to install it. Safari requires a bit more work with a plugin installer & needing to go to the extension "store," if Apple will even permit this.

Event handlers in the browser extension may be placed globally, in content scripts injected into each web page, or pushed via the content script to local DOM elements written by 3rd parties.

FLOW DIAGRAM: Kinect =====> Browser plugin/native code =====> Browser extension ===(Javascript+DOM events)==> Any web page

Note: As of now we are using OpenNI/NITE for tracking and gesture rec. Download the precompiled libs from:


Right now we only support Macs. All dependencies are statically compiled and in the repo.

Linux likely automatically works, although we don't distribute pre-compiled dependencies for it (yet).

Windows seems wayyy complicated, and we are OS snobs. Perhaps once Mac & Linux are done we'll think about trying it. Outside code is always welcome ;)


SAFARI: Safari needs it's own browser plugin & browser extension. webkit-plugin-mac/ contains the plugin, and the extension is in safari-extension-mac/. Unfortunately it does not like soft links, so you must in your terminal run

cd safari-extension-mac/DepthJS.safariextension && ./
Build & run the Xcode project in webkit-plugin-mac, then once inside Safari, enable developer tools & extensions, and finally add the extension under safari-extension-mac/ in Extension Builder. If you click on Inspect Global Pages, you'll see output confirming if it could connect to the Kinect or not (it should be plugged in).

Safari is currently our active development browser because of XCode & GDB. It will work before Chrome at any given moment.

CHROME: Chrome extensions support native code, which needs to be compiled. It's under npapi_plugin/ Run to create it.

The chrome extension is located in chrome-extension-mac/.

Future work

In addition to the obvious improvements to our gesture recognition, we need to make the install process easier for end-users.

Eventually when it's ready it will be a 1-click install. As this is a side side project and some of us are trying to finish our PhDs cough "ontime," updates have been steady but infrequent.