Title: BlenderFace - A Package for Evaluating Rater-Free Video-Based
Measurement of Facial Movements

Authors: Rainer Alexandrowicz, Axel Zinkernagel, Tanja Lischetzke, Manfred Schmitt

Affiliation: Alpen-Adria Universitaet Klagenfurt

Abstract:
The current gold standard for assessing facial emotion expressions relies on the
commercial Facial Action Coding System (FACS), requiring certified trained
raters to laboriously scrutinize a recording with respect to more than 60
so-called Action Units (groups of muscles inducing mimic expressions such as
Lower Lip Depressors or Chin Raisers, for example). We present a new rater-free
facial expression assessment system allowing for arbitrarily fine graded
automated capturing of mimic expressions using solely open source software. We
present the assembly along with an R package allowing for processing the output.
In an example evaluation we show how true and fake facial expressions of
emotions like anger or disgust might be differenciated with the recorded
parameters.