ExpressionApp is a sample application using the AR SDK to extract face expression

signals from video. These signals are

used to control the expressions, pose and gaze of a 3D morphable face model. The
application can either process
real-time video from a webcam or offline videos from files. It illustrates the
facial keypoints that are tracked, plots
the expression signals that are derived, and renders an animated 3D avatar mesh.

The application runs either the Face3DReconstruction, or FaceExpression feature,

depending on which expression mode is
used. The expression mode is toggled using the '1' and '2' keys on the keyboard
1 - Face3DReconstruction expression estimation
2 - FaceExpression expression estimation (default, and recommended for avatar
animation)

The FaceExpression mode is preferred for avatar animation. Note that

Face3DReconstruction is demonstrated for its
ability to track the face over time for AR effects. This feature enables identity
face shape estimation on top of
expression estimation and is better demonstrated in the FaceTrack sample
application. The resulting expression weights
from FaceExpression is more accurate than from Face3DReconstruction.

For details on command line arguments, execute ExpressionApp.exe --help.

For more controls and configurations of the sample app, including expression
definition and conversion to ARKit
blendshapes, please read the SDK programming guide. It also contains information
about how to control the GUI which can
be enabled by running the application with the --show_ui argument.