Microsoft’s Xbox Kinect, Nintendo Wii and Sony’s PlayStation Move all use motion-sensing game controllers to provide an added dimension of user engagement and “fun” to the game play of formerly thumb-bound fans, forced to use a hand-held game controller. Apple’s iPhone ushered in the widespread use of gesture controls, such as swiping or pinching and pulling a screen, as well as shaking the device to initiate software behaviors.
Now, forward looking web developers are moving beyond the mouse and keyboard to embrace innovative, sometimes “hands free” methods, of providing full user control over the computing and display environments.
Now, forward looking web developers are moving beyond the mouse and keyboard to embrace innovative, sometimes “hands free” methods, of providing full user control over the computing and display environments.
If this seems like the kind of cutting edge component that you would like to add to your website, then headtrackr (www.github.com/auduno/headtrackr/), a JavaScript library for real-time face tracking and head tracking via webcam, provides unique possibilities worthy of exploration.
Links from the depository provide additional information on the algorithms and tools, as well as the science behind the technology, providing for interesting further reading, such as the guide at https://dev.opera.com/articles/view/head-tracking-with-webrtc/.
The headtrackr script uses the WebRTC/getUserMedia standard, supporting browsers able to stream video and audio content directly from native devices, such as a webcam.
According to developer Audun Mathias Øygard, one rather exciting use for WebRTC is head tracking — detecting the movement of the user’s head (or other appendages) in relation to a webcam — which in turn allows developers to create gesture-based controls.
Øygard cites FaceKat (www.shinydemos.com/facekat/), an Opera 12 release demo employing head tracking as an example of the technology, noting that Opera 12 was the first desktop browser to support camera access via the getUserMedia API. There is also a video showcasing the technology at https://player.vimeo.com/video/44049736 that might give you an idea of its potential.
The demo and actual production examples are best when used with a laptop that has a builtin camera and a browser that has camera webRTC/getUserMedia support (you can see the overview of browsers supporting getUserMedia at https://caniuse.com/stream).
Better results are obtained when the user’s face is evenly lighted and looking ahead.
Using this technology is as simple as downloading the minified headtrackr.js library and including it on a webpage, just as with any other JavaScript, i.e. using HTML 5:
<script src=?headtrackr.js?></script>
To implement the script, the following code initiates the headtrackr using a video element to contain the mediastream and a canvas element to copy the videoframes to:
<canvas id=?inputCanvas? width=?320? height=?240? style=?display:none?></canvas>
<video id=?inputVideo? autoplay loop></video>
<script>
var videoInput = document.getElementById(?inputVideo?);
var canvasInput = document.getElementById(?inputCanvas?);
var htracker = new headtrackr.Tracker();
htracker.init(videoInput, canvasInput);
htracker.start();
</script>
Once the headtracker script is initiated, it will regularly generate the JavaScript events headtrackingEvent and facetrackingEvent on the relevant document, allowing coders to program various interactions between the site and its visitors, based on where they look.
“The event headtrackingEvent has the attributes x, y, z, which tells us the estimated position of the users head in relation to the center of the screen, in centimeters,” Øygard explains. “The event facetrackingEvent has the attributes x, y, width, height and angle, which tell us the estimated position and size of the face on the video.”
This allows users to either create an eventlistener to handle these events, or if using the three.js script, one of the pre-packaged controllers from this library could be used to create pseudo-3D images, also known as “head-coupled perspective” effects.
