Cascade classifier based face detection modified from http://www.adapteva.com/white-papers/face-detection-using-the-epiphany-multicore-processor to fit the parallella board and current eSDK.
Image: ubuntu-15.04-esdk-2016.3-headless-z7020
eSDK: 2016.3
sudo apt-get install libopencv-dev
cd code
./buils.sh
cd code/release
run.sh
"{ i | input | | Input image file }"
"{ c | classifier | lbpcascade_frontalface.xml }"
"{ g | grouping | 3 | Number of detections in group }"
"{ o | output | | Output filename }"
"{ h | host | 0 | Run detection on Epiphany | 1 | Run detection on ARM }"
"{ n | numcores | 16 | Number of working cores }"
"{ l | log | | Name of log-file }"
example:
./EpFaceHost i g20.jpg c lbpcascade_frontalface.xml g 3 o t1.jpg h 0 n 12 l 1.log
The green circle is Epiphany classify result, the red rectangle is opencv classify result
Host code
Launch device program in c/ep_cascade_detector.c, function ep_detect_multi_scale_device.
Device code