Face Tracking / Expression Recognition

From the age of Face Detection, we have come to an age where Face Tracking is feasible without a professional setup. All you need is a decent machine with a Kinect. If you have tried / tinkered with ofxFaceTracker, faceAPI or FaceDetect, I am sure you will love this. FaceShift can now be downloaded from their website. Kyle has developed a ofxaddon ofxfaceshift.

My observations: The tracking is very accurate and was impressed with the way it tracks the blinks and eye ball movements. Frame rate is good and the tracking does not have any gittering.

All this said, it comes with lot of limitations. Most of them comes with kinect. Like the minimum distance where the user has to stand / sit. Lighting conditions etc. and you have to train the software with few poses for good tracking results. So still its far from usage in realtime setups. But as  of now can be used for animation, expression tracking etc.


Tried playing around with openCV blobs and created a palette. This is my first project in openframeworks. I had this idea for the benetton’s window project while I was in fabrica but did not implement it then. As the logic is simple, thought of trying it in openframworks.

Basically it detects the consumers (users) walking by the window and generates a palette with predefined colors. The idea is to have the color pallete of the collection in the showroom. I still want to make the appearance of each bar little smooth.

Play Pong using Mouse, Keyboard, Webcam, Wii, iPhone and Kinect

This is a part of another project where we were interested in comparing different modes of interface for the game of Pong. The game is built in such a way that the user can choose either Mouse / Keyboard / Wii / iPhone / Webcam (color tracking) / Microsoft Kinect to play the game. This video shows the game being played using kinect.

Thanks to Suvani Suri for helping me to document this and also Manasi for her amazing tripod.