I have been hard at work getting the face tracking system ported over to the Raspberry Pi 2 in an organized fashion, and I am definitely just about there.
Here is a teaser video showing off the simplest working version.
I even put forth the initial efforts towards a full fledged C++ API! I now have PWM and Servo control very well supported for the PiSoC so that it can be very easily embedded into your C++ programs.
Anyway, the tracking system shown in the video is pretty good, but I’m still working on it. I am getting about 10fps capturing at 960×720 in full color for face detection. But you might notice I am only getting about 25% usage out of the cpu! That’s where the four cores of the new Raspberry Pi2 come in- I am only pushing one core to its limits, I need more!
So with a little extra work this morning dedicated to multithreading the detection and capture processes, I have already nearly doubled the performance.