To differentiate between two different faces, you will need to use the Cognitive Face skill. The Face Tracking skill in ARC is primarily used for detecting and tracking faces, but it does not have the capability to differentiate between individual faces.

Here's a brief overview of how you can achieve face differentiation using the Cognitive Face skill:

1. Set Up Cognitive Face Skill:

   • Ensure that your Cognitive Face skill is properly configured and connected to the internet, as it uses cloud services to recognize and differentiate faces.
   • Follow the instructions in the Cognitive Face skill manual to teach the system to recognize different faces. This involves capturing and uploading images of the faces you want to differentiate.

2. Teach the System to Recognize Faces:

   • Use the Learn function in the Cognitive Face skill to teach the system the faces you want to recognize. This will store the facial features in a global cloud database.

3. Detect and Differentiate Faces:

   • Use the Detect function in the Cognitive Face skill to recognize faces in real-time. The skill will return the name, estimated age, and emotional status of the detected face.
   • You can use the detected information to differentiate between the faces and trigger specific actions or responses in your robot.

4. Integrate with Face Tracking:

   • While the Cognitive Face skill handles the differentiation, you can still use the Face Tracking skill to keep the camera focused on the face. This way, the Cognitive Face skill can continuously analyze the face in view.

Example Workflow:

1. Add and Configure Skills:

   • Add the Cognitive Face skill to your ARC project.
   • Add the Face Tracking skill to your ARC project.

2. Teach Faces:

   • Use the Cognitive Face skill to teach the system the faces you want to recognize. This can be done by capturing images and associating them with names.

3. Detect Faces:

   • Use the Cognitive Face skill to detect faces in real-time. The skill will return the name and other details of the detected face.

4. Track Faces:

   • Use the Face Tracking skill to keep the camera focused on the face. This ensures that the Cognitive Face skill has a clear view of the face for recognition.

Example Script:

Here is an example of how you might use the Cognitive Face skill in a script to recognize and differentiate faces:

// Start face tracking
ControlCommand("Face Tracking", "StartTracking");

// Detect face using Cognitive Face skill
ControlCommand("Cognitive Face", "Detect");

// Get the detected name
var detectedName = getVar("$CognitiveFaceName");

// Check the detected name and perform actions
if (detectedName == "Person1") {
    print("Hello, Person1!");
    // Add actions for Person1
} else if (detectedName == "Person2") {
    print("Hello, Person2!");
    // Add actions for Person2
} else {
    print("Face not recognized.");
}

Resources:

• Cognitive Face Skill Manual
• Face Tracking Skill Manual
• Omron HVC-P Skill Manual

By using the Cognitive Face skill, you can effectively differentiate between different faces and trigger specific actions based on the recognized face. If you have any further questions or need additional assistance, feel free to ask!

Cognitive face is/was the technology for exactly that.

however, due to privacy and AI concerns of protesting groups, the technology has been disabled by Microsoft.

Synthiam participated in interviews about our use case to continue providing access to the cognitive face service. However, at some point a year ago, Microsoft disabled it without notice. We sent dozens of emails to every contact from the AI Ethics group that we could find, but no one responded.

The robot skill remains in the list - but the service is still disabled by Microsoft. We’re still hoping that someday they realize their error and restore access to it.

in the meantime, it’s a significant letdown. Specifically for education users and students. Because experiencing the technology is exactly what is needed to understand the dangers.

so, in the meantime you can use Object tracking by training the faces as different objects. It’s not as reliable and a little clunky , but it will work (sort of)