ros_web_apis

Some WEB APIs for ROS usage

ms_face_api

This is a ROS wrapper for the Microsoft Cognitive Services FACE API. It offers ROS services to detect faces, extract face markers, and recognise attributes like "gender", "smile", and "age". Also, faces can be trained for identification and if a face is trained, the detection will also contain the identity of a person in the image.

Services:

Detection and optional Identification

• call service cognitivefaceapi/detect to detect faces in an image
• Service Definition: ms_face_api/Detect
  • An image can either be defined by defining in the requests
    1. a sensor_msgs/Image,
    2. a filename, or
    3. a topic where the image is read from (waiting for a timeout)
  • It returns an ms_face_api/Faces data structure
  • if the identify flag is true in the request, the API tries to identify all the detected faces

Face Training

• call service cognitivefaceapi/add_face to add the largest face in the provided image with a given name
• Service Definition: ms_face_api/AddIdentityFace
  • An image is defined in the same way as for Detect (se above)
• The person is given a name in the request in the person field. If this person is already known, then the image is added to the training set for that person. If the person is not known, a new person is created and the face added to the training set.
• This service immediately retrains the face identifications system.
• Afterwards, the trained person should be identified by the Detect service (potentially after some delay as training can take some time)