Affect valence inference from facial action unit spectrograms

The face provides an important channel for communicat-ing affect valence, the positive or negative emotional charge of an experience. This paper addresses the challenging pat-tern recognition problem of assigning affect valence labels (positive, negative or neutral) to facial action sequences ob-tained from unsegmented videos coded using the Facial Ac-tion Coding System (FACS). The data were obtained from viewers watching eight short movies with each second of video labeled with self-reported valence and hand coded us-ing FACS. We identify the most frequently occurring Facial Actions and propose the usefulness of a Facial Action Unit Spectrogram. We compare both generative and discrimina-tive classifiers on accuracy and computational complexity