This ROS package provides a node which takes subscribes to an image and uses the Intel Neural Compute Stick 2 to detect human poses. These are published and a debug image is provided.
Expects Image message on "/camera/image_proc"
Publishes self definded "HumanPoseEstimation" message on "/poses"
Publishes debug image on "/human_pose_debug"
bool "publish_debug_image": set false to disable debug image (improves performace)
float "score_threshold": poses with a score less than threshold are not published (default is 150)
You need to install OpenVino and its InferenceEngine. Make sure to source the provided "setupvars.sh". For more information see section "Build the Demo Applications" here:
https://docs.openvinotoolkit.org/latest/omz_demos_README.html
The code is based on the OpenVino model zoo. More information can be found here:
https://github.com/openvinotoolkit/open_model_zoo/tree/master/demos/human_pose_estimation_demo
More information about the model can be found here: