ros-object-detection-haar

Object detection package for ROS that uses Haar cascade classifiers. Requires OpenCV. Useful for simple face detection, cat detection, etc. on crappy embedded hardware. You can get 15 fps+ on a Raspberry Pi.

Try it

rosrun object_detection_haar object_detection_haar_node __ns:=/camera

Parameters:

• cascade (string) -- name of the cascade to use. Must be an OpenCV-compatible Haar cascade in the nodes/haar directory.
• topic_image (string) -- topic to listen for images. Defaults to "image_raw".
• topic_detections (string) -- topic to output semantic predictions. Defaults to "detections". Outputs a JSON string which is a list of [[x0,y0,w0,h0], [x1,y1,w1,h1], ...].

Subscribers:

• image_raw (sensor_msgs/Image)

Publishers:

• detections (std_msgs/String)

Disclaimer

This is provided "as-is" for educational purposes. I am not liable for any damage or injury that may result from the use of this software.