Bob ROS Gestures

Hand and face detection code to use for Bob ROS. Code is based off of ROS 2 but has been translated into ROS.

Creators (ask us if you need help): Evan Conway (original ROS 2 code) and Andrew Vithoulkas (conversion to ROS and continued development)

Core Files

• publisher.py: Publishes camera to a ROS topic (parameters: source)
  • Parameters:
    • source (int): what camera source to publish (defaults to 0)
  • Topics:
    • Output:
      • /camera_[x] (where [x] = source)
• subscriber.py: Reads camera data from topic and scans for hands / faces depending on the specified parameters
  • Parameters:
    • camera_topic (string): the camera topic to listen to (defaults to "camera_0")
    • display (bool): whether or not to visually display the results (defaults to false)
    • detect_faces (bool): toggles basic face detection (defaults to false)
    • landmark_faces (bool): toggles face landmarking (defaults to false)
    • landmark_hands (bool): toggles hand landmarking (defaults to false)
  • Topics:
    • Input:
      • /[camera_topic] (depends on what the parameter camera_topic is)
    • Output:
      • /face_detection_results
      • /face_landmark_results
      • /hand_landmark_results

Model Files:

• blaze_face_short_range.tflite: Data for short range face detection
• face_landmarker.task: Data for face landmarking
• hand_landmarker.task: Data for hand landmarking