sb_facedetect_demo

README

face detection demo on SensorBee

With OpenCV cascade classifier function. This demo is putting a mask on faces. Target OS is Linux or Mac.

put mask on faces demo

prepare target file and model file

Prepare target video file, mask file and cascade model file. Model files, for example, opencv/data can use.

Below tree is a sample for data set

./data
├ target_video_file.avi
├ haarcascade_frontalcatface.xml
└ mask.png

edit BQL

cp rectangle_mount.bql.sample rectangle_mount.bql

Edit file paths at rectangle_mount.bql L2, L7, L36.

CREATE PAUSED SOURCE camera TYPE opencv_capture_from_uri WITH
    uri="data/target_video_file.avi",
    frame_skip=0, next_frame_error=false
;

CREATE STATE face_detector TYPE opencv_cascade_classifier WITH
    file="data/haarcascade_frontalcatface.xml"
;

...

CREATE STATE mount TYPE opencv_shared_image WITH
    file="data/mask.png"
;

build SensorBee and make working directory

$ build_sensorbee --download-plugins=false
$ mkdir result

Will be made "sensorbee_main.go" and "sensorbee" (binary file). If fail to build by not fund libraries, download each library by go get.

execute the BQL

$ ./sensorbee runfile -c conf.yaml rectangle_mount.bql

Will be created "detected.avi" and "mounted.avi" on "result" directory.

put mask on a certain person

TODO