Simple ROS face detection package using traditional computer vision techniques.
Author: Tasuku Miura
The package uses HAAR_CASCADE based models. The specific model can be specified
in the launch file after placing the respective model in the /models folder.
This can be used with any client side that publishes raw images. In this
particular package I chose to use video_stream_opencv running on a Raspberry
Pi 3. See ROS+RaspberryPi Camera Module #3: An alternative package for
publishing images from
Raspi.
for specifics.
To run, launch webcam.launch from the raspberry pi.
Once you have confirmed that the topics are being published (check for
/webcam/image_raw/), then launch the face_detect.launch found in the
launch directory in the ros_face_detect package.
$ roslaunch face_detect.launch
If you do not want to output bounding rectangles for performance reasons, go to the face_detect.launch file
and set the param with name bound_rectangles to false.
Note: As this was intended to run on Jetson TX1, I have set the default as the cuda implementation,
thus CMakefile.txt and face_detect_node.cpp reflect this. If you would like to run the CPU version
make the necessary edits to CMakefile.txt by changing the names in add_executables() and also the header
file included in face_detect_node.cpp to #include face_detect_cuda.h.