Other than bookkeeping operations, such as setting the appropriate tracking state and incrementing the tracking time, the core of the tracking algorithm is the multi-level fitting procedure, which is highlighted in the preceding code snippet. The fitting algorithm, implemented in the face_tracker::fit function, is applied multiple times with the different search window sizes stored in face_tracker_
params::ssize, where the output of the previous stage is used as input to the next.In its simplest setting, the face_tracker_params::ssize function performs the facial feature detection around the current estimate of the shape in the image: